The Brain Injury Guidelines: Can we avoid talking to neurosurgeons?

One of the biggest headaches in modern medicine is the apparent requirement to call busy specialists just to confirm what seems like an obvious treatment plan. In emergency medicine, this often happens when a patient needs to be admitted under one service, but with a medical problem related to another speciality. We get stuck in the middle, tasked with pointless phone calls, and often take the brunt of a frustrated specialist. The classic example is needing to admit a patient with a small brain bleed for observation and a repeat CT scan, but having the admitting team ask us to call neurosurgery to confirm the obvious plan. Neurosurgery is (probably rightly) frustrated by the volume of these calls for clearly non-surgical patients who really don’t need their expertise. Wouldn’t it be nice if we could just skip these calls?

The YouTube Video

The paper

Joseph B, Obaid O, Dultz L, Black G, Campbell M, Berndtson AE, Costantini T, Kerwin A, Skarupa D, Burruss S, Delgado L, Gomez M, Mederos DR, Winfield R, Cullinane D; AAST BIG Multi-institutional Study Group. Validating the Brain Injury Guidelines: Results of an American Association for the Surgery of Trauma prospective multi-institutional trial. J Trauma Acute Care Surg. 2022 Aug 1;93(2):157-165. doi: 10.1097/TA.0000000000003554. PMID: 35343931

The Methods

This is a multicenter prospective observational cohort from 10 level I and level II trauma centers in the United States. 

Patients

They included adult (16 and older) trauma patients with traumatic brain injury and positive CT scans (bleed or skull fracture). They excluded patients transferred from outside institutions, requiring emergent surgical intervention, and those with missing information. 

Intervention

They looked at the brain injury guidelines (BIG), which classify patients as follows:

You are always classified in the highest possible group. (In other words, you need to satisfy 100% of the listed criteria to count as BIG 1 or BIG 2).

Outcome

Their primary outcome of interest was need for neurosurgical intervention.

The Results

They screened 2432 patients, and their final cohort consists of 2033 patients. 15% were classified as low risk (BIG 1), 15% as medium risk (BIG 2), and 70% as high risk (BIG 3). There was excellent agreement between clinically assigned and researcher verified BIG category, with a kappa of 0.992. Overall, almost 20% of the patients received blood transfusions. (In other words, these are not just isolated head injury patients, but truly multisystem trauma patients.) 

The initial head CT findings are as below:

No BIG1 patients had neurologic deterioration, and none required neurosurgical intervention. That is good news, but unfortunately this paper presents no stats at all, we have no idea how high the 95% confidence interval might extend (but given that the BIG1 cohort only had 301 patients, it is probably pretty high.) There were 1.3% of patients in the BIG group that had worsening of their bleed on repeat CT scan. 1 patient in this cohort had a readmission within 30 days, but they say it wasn’t related to traumatic brain injury. 

Among BIG 2 patients, only 0.7% had clinical deterioration, and although 7% had progression of the bleed on repeat CT, none of these patients required neurosurgical intervention. (Again, it is the upper limit of the 95% confidence interval that is really going to matter for practice change.)

For BIG 3 patients, 16% deteriorated clinically, 21% had progression on repeat CT scan, and 20% had neurosurgical intervention.

They state that implementation of the BIG guidelines would have resulted in 425 fewer repeat CTs, 401 fewer prolonged hospitalizations, and 511 fewer neurosurgical consultations in this cohort of 2000 patients. 

My thoughts

I love the goals of this study, but unfortunately the data is not very strong. Too many patients were lost to follow-up, or excluded because of missing data, and even one or two bad outcomes in that group would dramatically alter the results. Just the fact that one or two outcomes could dramatically alter the results tells you a lot about the certainty of the data, and the fact that this data is presented without any statistics (specifically the 95% confidence intervals we need) is troubling to me. Overall, this study is not enough to verify the guidelines as they are presented, but that doesn’t mean we can’t learn from them and adjust our clinical practice. 

If you are going to try to apply these results, I think it is going to be important to consider selection / referral bias. Not very many patients were categorized as low risk (which sort of surprises me based on my experience). I think this likely represents referral bias. This study takes place at trauma centers, and therefore skews towards the sickest trauma patients. In the community, the majority of the patients I see have relatively minor brain bleeds, and so implementation of these guidelines probably has a much bigger potential for benefit.

On the other hand, we need to be careful about generalizability. These patients were all monitored in a trauma center. They were admitted under trauma surgeons. So even if neurosurgery was never involved, they were cared for by teams very experienced in trauma. The results seen here may not apply to patients admitted to a small rural hospital where a single nurse with limited trauma experience covers the entire ward overnight.

The results probably look worse than they would in isolated head injury patients in the average emergency department, given that these were generally multi-system trauma patients presenting to level 1 trauma centers. However, there were a large number of patients (399 as compared to the 2,033 included) who were excluded because of missing information or loss to follow-up. This adds tremendous uncertainty to the data, and if you assume patients with bad outcomes might be more likely to be lost, this suggests the real world outcomes might be worse than seen here.

The data here looks even worse when you read “five centers had greater than 30% compliance with the BIG with regard to routine repeat head CT imaging and neurosurgical consultation.” The threshold they used for high compliance was 30%!! In other words, not even the big trauma centers trust these guidelines yet, which probably indicates they aren’t ready for the average hospital.

It is reassuring that no patients in the BIG1 group had clinical deterioration or required neurosurgical intervention. Those are the outcomes that matter if we are going to manage these patients independently. However, the 1% rate of worsened bleeding, in a study with imperfect methods and without a calculated 95% confidence interval, is not perfect considering their proposed plan to discharge these patients home from the emergency department. This is especially true as repeat CT was not universal. (According to these guidelines BIG1 and BIG2 patients were not supposed to get repeat CT scans. Only about 70% of these groups had a repeat CT scan.) Therefore, there are likely more patients with CT progression that simply weren’t identified. 

In the long run, I think we are going to need more science focusing on the individual components of these guidelines. For example, they make every patient with any form of antiplatelet medication a BIG3 or high risk patient, mandating a neurosurgical consultation. I am not sure that makes sense. Those patients definitely need admission, observation, and a repeat head CT, but I am not sure adding an automatic neurosurgical consultation adds much for that population. 

I really appreciate the goal of this study. Reflex neurosurgical consultation helps nobody. Automatic repeat CT for everyone is probably unnecessary. We really need better guidance around resource utilization for this group of patients. This study is far from perfect, but I hope we see more data supporting practice change in this area. 

Personally, there are already a lot of patients I would prefer to simply admit for observation without calling neurosurgery. If I was left to my own devices, I would probably use these guidelines to help guide my judgment in that decision. Unfortunately, as things currently stand, I am not allowed to make that decision. When I talk to my colleagues about getting these patients admitted, I am met with the universal refrain of, “we will not admit until you talk to neurosurgery.” Thus, until we get buy-in throughout the hospital, my hands are tied. 

Bottom line

In the first prospective look and the brain injury guidelines, no patients classified as low risk required neurosurgical intervention. This might help us limit resource utilization, by decreasing our use of routine repeat CT and neurosurgical consultation for low risk patients. However, this data has a number of issues, and probably is not strong enough on its own to justify complete implementation of these guidelines. 

Other FOAMed

Evidence based medicine is easy

The EBM bibliography

Evidence based medicine resources

EBM deep dives

References

Joseph B, Obaid O, Dultz L, Black G, Campbell M, Berndtson AE, Costantini T, Kerwin A, Skarupa D, Burruss S, Delgado L, Gomez M, Mederos DR, Winfield R, Cullinane D; AAST BIG Multi-institutional Study Group. Validating the Brain Injury Guidelines: Results of an American Association for the Surgery of Trauma prospective multi-institutional trial. J Trauma Acute Care Surg. 2022 Aug 1;93(2):157-165. doi: 10.1097/TA.0000000000003554. PMID: 35343931