Without evidence of benefit, an intervention should not be presumed to be beneficial or safe.

- Rogue Medic

Prehospital Intravenous Fluid Administration is Associated With Higher Mortality in Trauma Patients – Part III

ResearchBlogging.org

Also posted over at Paramedicine 101 (now at EMS Blogs) and at Research Blogging. Go check out the rest of the excellent material at these sites.

This paper is discussed in Prehospital administration of IV fluids to trauma patients: EMS Research Episode 5.

This is also discussed by Dr. Chris Russi in the podcast Russi’s Research Review – Episode Two (Fluid Administration in Traumatic Injury Patients).

Continuing from Part I and Part II.

The 5 most common EMS procedures as documented in the NTDB (National Trauma Data Bank®) are listed in this table.


Click on images to make them larger.
 

In Part II, I explained the problems with the NTDB claiming that only 49.3% of trauma patients had IV fluids starts documented. While that should raise questions, looking at the data on the rest of the top 5 procedures makes the questions even more obvious.

Looking at the abridged Table 1 that I included in Part II, to show the problem with the number of IVs documented, unit12medic recognized the problems with the rest of the data. I changed the abridged Table 1 only changing what I underlined. I removed the underlining from the IV fluids and added underlining to the other procedures that are documented with unrealistic frequency.

Only 8.1% had spinal immobilization?

IV (IntraVenous) starts were over 6 times more common than spinal immobilization for these trauma patients?

Is there any place where this is the way EMS does things?

8.1% had needle decompression, but only 4.4% were hypotensive?

Is there any reason to assume that a patient who is not hypotensive will improve by having a large needle stuck in his chest?

Chest decompression is almost 3 times more common than intubation in these trauma patients?

More than one in every 12 1/2 trauma patients had chest decompression.

MAST application is 3 1/2 times more common than spinal immobilization?


Click on images to make them larger.

When broken down by procedure among those who did not have an IV start documented vs. those who did have an IV start documented, things become even more odd. The authors claim to have adjusted for all of these variables, but the difference in rate of application goes from small to what appears to be inexplicably huge. I can’t explain this except if the data do not reflect reality.

A total of 776,734 patients with complete prehospital procedure files were identified from the 1,466,887 total patients in the National Trauma Data Bank.[1]

Multivariable logistic regression was used to examine the relationship between prehospital IV and mortality in the 311,071 patients with complete data.[1]

1,466,887 total NTDB patients.

776,734 patients with complete prehospital procedure files (53% of 1,466,887).

311,071 patients with complete data (40% of 776,734 and 21% of 1,466,887).

What would result in such a dramatic difference between MAST with an IV Start and MAST without an IV Start?

53.4% is 281 times more common than 0.19%. What could possibly explain this? Differences in protocols?

Even though the authors concluded that IV Starts produced worse outcomes, the procedure most strongly correlated with IV Starts was determined to have produced a protective effect. Most IV Start patients had MAST applied, while less than one fifth of one percent of the No IV Start patients had MAST applied.

Intubation (OR 1.57) and spinal immobilization (OR 1.42) were found to increase the odds of death by much more than IV Starts (OR 1.11) were increasing the odds of death.

If these numbers were valid, the increased odds of death should result in strongly worded warning letters on the hazards of spinal immobilization and intubation of trauma patients.

The less frequent the procedure/condition, the larger the calculated increased risk of death. This also means that the larger the calculated increased risk of death, the larger the confidence interval.

With the unbelievably low rate of spinal immobilization, which is more likely the opposite of what is recorded, should we trust any of the numbers from the NTDB?

If we cannot trust any of the numbers from the NTDB, we must doubt the least frequent numbers.

We did not adjust for cardiopulmonary resuscitation because the data on cardiopulmonary resuscitation appeared to be biologically implausible:[1]

 

I read this and thought that they were going to explain that compressions of the chest are pointless, when there is no blood in the vessels to circulate. And this is true.

I thought that they might also explain that compressions of the chest are pointless, when there is a complete obstruction to circulation. And this is true with arrests due to pulmonary embolus or cardiac tamponade.

But that was not their point.
 

the mean systolic blood pressure of penetrating trauma patients who received cardiopulmonary resuscitation was 118 mmHg.[1]

 

That is higher than my blood pressure.

I am pretty sure that the AHA (American Heart Association) does not want anybody using CPR (CardioPulmonary Resuscitation) on people with good blood pressures. CPR is for people with no blood pressure (or for children with extremely low blood pressures).

Does this give us a hint about the reliability of the information used?

I think so.

Regardless of the problems with the data in this study, we still have no evidence that giving fluids to patients before bleeding is controlled improves outcomes.

Regardless of the problems with the data in this study, we still have no evidence that giving fluids to patients before bleeding is controlled is safe.

I generally agree with the conclusion, that fluids should not be given (I would add – before bleeding is controlled), but I do not think that this study provides valid evidence to support that conclusion.

I am adding the article below to Part I. This is the reporting by Medscape. This was not put on line after I wrote about it, but I missed it when I originally wrote this. Medscape is a web site edited by doctors, but even they did not seem to notice the flaws of this study.

From Medscape Medical News
Prehospital IV Fluids May Be Harmful for Trauma Victims
Medscape
Laurie Barclay, MD
January 20, 2011
Article

Footnotes:

[1] Prehospital Intravenous Fluid Administration is Associated With Higher Mortality in Trauma Patients: A National Trauma Data Bank Analysis.
Haut ER, Kalish BT, Cotton BA, Efron DT, Haider AH, Stevens KA, Kieninger AN, Cornwell EE 3rd, Chang DC.
Ann Surg. 2010 Dec 20. [Epub ahead of print]
PMID: 21178760 [PubMed – as supplied by publisher]

Full Text in PDF format from www.medicalscg.

Haut ER, Kalish BT, Cotton BA, Efron DT, Haider AH, Stevens KA, Kieninger AN, Cornwell EE 3rd, & Chang DC (2010). Prehospital Intravenous Fluid Administration is Associated With Higher Mortality in Trauma Patients: A National Trauma Data Bank Analysis. Annals of surgery PMID: 21178760

.

Comments

  1. Is there any reason at this point not to completely throw the NTDB out the window if this is the quality of data contained within?

    • usalsfyre,

      Is there any reason at this point not to completely throw the NTDB out the window if this is the quality of data contained within?

      This is a start. It has a long way to go.

      There do seem to be significant problems with the data, but I have not seen enough of the data, or how the data are collected, to make that kind of decision.

      I also need to write a correction for my post on Spine Immobilization in Penetrating Trauma: More Harm Than Good? This is also based on the NTDB. I looked at my notes on the PDF and I have a bunch of comments about the problems with the data. When I wrote the post, I did not include any of these negative comments. I let my biases affect what I wrote.

  2. This study was done in retrospect. That says to me they got to pick and chose which documentations went into this study. More studies should be done on this, but they need to go with reliable information.

    • just me,

      This study was done in retrospect. That says to me they got to pick and chose which documentations went into this study. More studies should be done on this, but they need to go with reliable information.

      There was really only one document used in this study. The information at the National Trauma Data Bank® is clearly not accurate, but there is research being done with this as the only data source.

      I agree. We definitely need to get reliable data.