Tag Archives: scoring

NFTI Scoring Revisited – Not Just For Triage Calculations?

Earlier this week, I wrote about a new tool for monitoring over- and under-triage for trauma programs. In place of using ISS as the metric for triggering review, the Need For Trauma Intervention (NFTI) is based on resource utilization during the initial portion of the hospital stay.

The original study was performed at a single Level I trauma center in Dallas. The authors then rolled it out as a multicenter study to test its overall reliability. However, the authors changed the focus in this work. The original paper focused on the development of a new tool to improve upon the evaluation of proper decisions to activate the trauma team. The authors have now extrapolated that their system predicts when a patient’s physiologic reserve is depleted. In turn, this should be the indicator that a trauma activation is needed.

The authors performed a convenience sample of 38 trauma centers around the US. Of these, 25 were adult only, 3, pediatric only, and 10 were combined adult/peds centers. Two years of data were collected from each. Injury severity score (ISS) and revised trauma score (RTS) were calculated for all patients. Outcomes analyzed were discharge location (home vs ongoing care), complications, and length of stay.

A complicated statistical model was adopted that evaluated the associations between higher ISS (> 15), lower RTS (< 7.84) and any positive NFTI factor. To refresh your memory, here’s the list of NFTI factors:

  • blood transfusion within 4 hours of arrival
  • discharge from ED to OR within 90 minutes of arrival
  • discharge from ED to interventional radiology (IR)
  • discharge from ED to ICU AND ICU length of stay at least 3 days
  • require mechanical ventilation during the first 3 days, excluding anesthesia
  • death within 60 hours of arrival

Here are the factoids regarding the new study:

  • Nearly 90,000 patient encounters were submitted over a 2 year period
  • The risk of experiencing a complication increased by 9x if NFTI+, 6x for ISS>15, and 5x for RTS<7.84
  • Odds of discharge to a continuing care facility was about 2.5x more likely if any of the three thresholds were met
  • Length of stay was significantly better predicted by NFTI

The authors conclude that NFTI was a better indicator of major trauma when compared to ISS and RTS. They claim that it is the best single definition because the model fit is better and that it has stronger associations with complications, discharge location, and length of stay.

Bottom line: Hmm, I’m not so sure. It’s a great idea and does allow us to drill down on those patients most in need of high-level trauma center resources. The authors admit that each tool (ISS, RTS, and NFTI) identifies some important patients that the others do not. It just seems that more of them tend to be identified by NFTI.

I always worry when complicated statistical models are needed to show these differences. This is a complex concept, so more sophisticated models may indeed be needed by virtue of the data that needs to be analyzed. Overtriage can be easily identified in many cases when NFTI- patients trigger a full trauma activation. Obvious undertriage occurs in NFTI+ patients with no activation.

But NFTI still does not obviate the need to search harder for undertriage. What about the case of a stab to the chest in the “box” region, who does not end up with a cardiac injury or hemo/pneumothorax? They would be NFTI- but mechanism positive.

How do we learn from NFTI+ patients who did not have a trauma activation. Just like using the Cribari grid, we must look at each individual chart and ask two questions:

  1. Did this patient meet any of our highest level activation criteria? If so, it is frank undertriage.
  2. If not, do we need a new criterion to catch this in the future?

So NFTI is a somewhat improved version of the Cribari grid. Sure, it can predict complications better, as well as length of stay (which may be related). But not discharge location, as claimed. As for being an indicator of depleted patient reserve, I think that’s just speculation at this point. Both tools can be used to automatically generate data for review from the trauma registry. And both will have some false negatives and positives.

My recommendation: This paper provides an academic argument that NFTI is somewhat better than the Cribari method. Now it’s time to get practical. Some enterprising trauma centers need to do a study where they use both systems side by side. How many charts for review are generated by each? How many false negatives and positives are there? How much work (abstractor / registrar time) is needed to analyze and act on the results? This is the only way we can answer the question of which one is better in the real world.

Reference: Rethinking the definition of major trauma: The Need For Trauma Intervention outperforms Injury Severity Score and Revised Trauma Score in 38 adult and pediatric trauma centers. J Trauma publish ahead of print, 2019.

NFTI: A Nifty Tool To Replace The Cribari Grid?

In my last post, I reviewed the use of the Cribari grid for evaluating over- and under-triage at your trauma center.  This technique has been a mainstay for over a decade, but has its shortcomings. The most important one is that it relies only on the Injury Severity Score (ISS) to judge whether some type of mistriage occurred.  As you know, the ISS is usually calculated after discharge, so it can only be applied after the fact.

Two years ago, the group at Baylor University in Dallas sought to develop an alternate method of determining who needed a full trauma team activation. They chose resource utilization as their surrogate to select these cases. They reviewed 2.5 years of their own registry data (Level I center).  After several iterations, they settled on six “need for trauma intervention” (NFTI) criteria:

  • blood transfusion within 4 hours of arrival
  • discharge from ED to OR within 90 minutes of arrival
  • discharge from ED to interventional radiology (IR)
  • discharge from ED to ICU AND ICU length of stay at least 3 days
  • require mechanical ventilation during the first 3 days, excluding anesthesia
  • death within 60 hours of arrival

Patients who had at least one NFTI criterion were considered candidates for full trauma activation, and those who met none were not. Here are the factoids for this study:

  • There were a total of 2260 full trauma activations and 2348 partial activations during the study period (a little over 900 per year for each level)
  • Roughly 2/3 of full activations were NFTI +, and 1/3 were NFTI –
  • For partial activations, 1/4 were NFTI + and 3/4 were NFTI –
  • Only 13 of 561 deaths were NFTI – and all had DNR orders in place

The authors concluded that NFTI provides an assessment of both anatomy and physiology using only measures of early resource utilization. They believe that it self-adjusts for age, frailty, and comorbidities, and that it is a simple and effective tool for identifying major trauma patients.

Bottom line: This is an elegant attempt to improve upon the simple (yet admittedly flawed) Cribari matrix method for assessment of major trauma patient triage. It was thoughtfully designed and evaluated at this one center. The authors recognize that it is based on retrospective data, but so is the Cribari technique. 

I believe that it may be an adjunct to Cribari. The matrix identifies gross under- and over-triage, but still requires the trauma program to review the outliers to see if mistriage actually occurred. It is basically a “first pass” that seeks to over-identify potential problem patients.

NFTI is similar, but it focuses on those patients who really should have been a full trauma activation due to their early need for critical resources to deal with their injuries. But is it enough? In my next post, I’ll review the follow-on paper from this group as they apply it to multiple trauma centers. And I’ll add some final thoughts on the subject.

Reference: Asking a Better Question: Development and Evaluation of the Need For Trauma Intervention (NFTI) Metric as a Novel Indicator of Major Trauma. J Trauma Nursing 24(3):150-157, 2017.

Appropriateness Of Nonsurgical Admissions

U.S. Trauma Centers that are verified by the American College of Surgeons must track the rate of trauma admissions to nonsurgical services. This is particularly important if the percentage of nonsurgical admissions exceeds 10% of their total admissions. The center’s performance improvement processes can then determine if the admission was appropriate and whether or not the trauma service should request a consult or transfer.

But how should we judge the appropriateness of nonsurgical admissions? There is tremendous variability in presenting mechanism and patient comorbidities. And the number of patients with some need for nonsurgical attention continues to grow with the rapidly increasing number of elderly falls.

The group at Southside Hospital in Bay Shore NY initially tracked all nonsurgical admissions and evaluated each individually at their community Level II trauma center. They then created and implemented a scoring system in order to develop a set of objective criteria that would predict patients better served with trauma consultation or admission.

The scoring tool was based on some of the information in the Optimal Resource Document, but was still somewhat arbitrary. The authors added criteria that reflected their own institutional philosophy of care. They explain their rationale clearly in the manuscript. Here is the final tool:

Criteria Points
Age > 65 years 1
3 or more comorbidities 1
ISS < 10 1
Ground level fall 1
No ICU admission 1
No need for surgical intervention 1
No blood products given 1

The maximum number of points possible is 7, with higher scores suggesting appropriateness for nonsurgical admission. The authors chose scores of 3 and 4 as the “grey zone” where further investigation was necessary to determine if a medical admission was proper. Lower numbers required trauma service admission, and higher ones did not.

The authors then examined changes in the percent of nonsurgical admissions after implementation, as well as mortality, morbidity, and hospital length of stay.

Here are the factoids:

  • Nonsurgical admission rates had historically been greater than 10% and had peaked at 28% at the time of scoring system implementation
  • After implementation, the nonsurgical admission rate dropped to under 10 %, where it remained for most of the time. There were a few spikes into the 14-17% range.
  • Mortality was insignificantly higher on the trauma service (2.1% vs 1.2%) as were complications (6.1% vs 5.5%)
  • Length of stay was statistically significantly longer on nonsurgical services (6.2 VS 5.1 days)

Bottom line: Centers that admit a large number of elderly falls patients may benefit from adopting this quick screening tool to determine the appropriate service. Ideally, all trauma patients would be admitted to the trauma service, but this is not feasible from a personnel and resource standpoint. If the number of potential nonsurgical admissions is high, applying a scoring system like this can help streamline the decision regarding admitting service.

Patients with very low scores (1-2) are obviously only appropriate for a trauma service admission. Likewise, those with very high scores (5-7) could easily and appropriately be managed on a hospital medicine service. The in-betweeners need more scrutiny by trauma program PI personnel to determine which service to admit to. 

Most importantly, don’t feel compelled to use this exact scoring system or threshold. Every hospital has different resources and a unique patient population. Add or remove criteria that you believe are appropriate. Adjust the threshold for added scrutiny as you see fit. Doing so will help keep your trauma PI workflow manageable.

Reference: Nonsurgical admissions with traumatic injury: medical patients are trauma patients, too. J Trauma Nursing 25(3):192-195, 2018.

Geriatric Outcome Prediction From The P.A.L.LI.A.T.E Consortium

The continuing rise in geriatric trauma cases seen at trauma centers has necessitated the creation of new infrastructure for evaluating, treating, and assessing outcomes in injured elders. The ability to predict the likely outcome after trauma is extremely important in shaping the management of these patients after discussion with them and their families. Unfortunately, the tools we have for those prognostications are rather complicated, yet rudimentary.

The gold standard to date is TRISS, which combines physiologic data (revised Trauma Score) at the time of first encounter with anatomic injury information (Injury Severity Score). This allows the calculation of a validated probability of survival (Ps).

However, TRISS is unwieldy and frequently cannot be calculated due to missing data. A consortium was created to address these shortcomings. Of course, they chose a name with an unwieldy acronym: Prognostic Assessment of Life and LImitations After Trauma in the Elderly (PALLIATE).

This group developed the Geriatric Trauma Outcome Score (GTOS) in 2015. They recently published a study comparing GTOS with the gold standard TRISS. This could be important since GTOS is easier to calculate, with less opportunity for missing data since it relies only on age, ISS, and presence of blood transfusion.

They calculated outcomes of nearly 11,000 patients at three centers, and found that GTOS worked as well as TRISS. The major advantage was that GTOS requires only three variables:

GTOS = Age + (ISS x 2.5) + (22 if blood transfused in first 24 hours)

Then, just to make your head spin a little more, the GTO score value gets plugged into this logistic model equation:

Bottom line: GTOS is helpful in some ways, but not in others. It does allow calculation of the probability of survival in elderly patients as well as traditional methods, but with more readily available data points. 

However, it is just a probability. It may predict that someone like your patient has a 3% probability of survival, but it cannot tell specifically that your patient is in the 3% vs the 97%. The consortium was trying to make it easier and more objective for clinicians to discuss care plans with family. But this is not really the case. 

And a bigger problem is that it gives us no guidance as to quality of life or level of independence for those patients who will probably survive. These factors are, by far, the most important ones when having those hard discussion with patient and/or family. We still need a tool that will guide us on functional outcomes, not just life or death.

Related posts:

Reference: A comparison of prognosis calculators for geriatric trauma: A P.A.L.LI.A.T.E. consortium study. J Trauma, publish ahead of print DOI: 10.109, 2017.

AAST Revises Renal Injury Grading

Organ injury scaling was developed to give clinicians and researchers a common language for describing and studying the effects of trauma. The Organ Injury Scaling classification for kidney injuries was developed by the AAST in 1989. Over time, it was recognized that grades IV and V were somewhat confusing, and some injuries were not originally included. An updated grading system was published this month to correct these shortcomings.

Grades I, II, and III remain unchanged. Grades IV and V are updated as follows:

  • Grade IV – originally encompassed contained injuries to the main renal artery and vein, and collecting system injuries. Revision: adds segmental arterial and venous injury, and laceration to the renal pelvis or ureteropelvic junction. Multiple lacerations into the collecting system used to be considered a shattered kidney (Grade V), but now remains Grade IV.
  • Grade V – orignally included main renal artery or vein laceration or avulsion, and multiple collecting system lacerations (shattered kidney). The revised classification includes only vascular injury (arterial or venous) and includes laceration, avulsion or thrombosis.

Reference: Revision of current American Association for the Surgery of Trauma renal injury grading system. J Trauma 70(1):35-37, 2011.