Does serum lactate predict death in drug overdose?
August 23, 2010, 9:44 am
Utility of serum lactate to predict drug-overdose fatality. Manini AF et al. Clin Toxicol 2010 Aug 12. [Epub ahead of print]
This derivation study was designed to evaluate the usefulness of serum lactate in identifying acute drug overdose patients at increased risk of death. The authors compared 50 poison-related fatalities (collected from cases reported to a regional poison control center over a 7-year period) with 100 controls (acute non-fatal drug overdose patients seen in 2 emergency departments over a 1-year period). All cases and controls had at least 1 serum lactate drawn in the emergency department; only the initial lactate was used for analysis.
The authors concluded that their data indicated a lactate level of 3.0 mmol/L was the optimal cutoff (sensitivity 84%, specificity 75% for predicting subsequent fatality). They state that their paper describes a retrospective derivation study that should be validated prospectively.
I am afraid, alas, that this prospective validation will never happen. Overdose fatalities are so infrequent that such a study would have to involve massive numbers of patients collected from many different emergency departments. This would involve so many variables — specific toxins, time of presentation, time lactate specimen was drawn, etc etc — that it would be impossible to determine if lactate itself was an independent or useful indicator.
In addition, there are many problems with this initial derivation study. Most importantly, the cases and controls were really two different sets of patients. The cases had a much higher incidence of baseline comorbidities: twice the rate of hypertension, coronary artery disease, and congestive heart failure, as well as three times the rate of diabetes. The cases came from all area hospitals with access to poison center consultation (not just two tertiary centers) over a 7-year period (not just one year like the controls). Furthermore, it was not possible to standardize time that the lactate specimen was obtained — it is entirely conceivable that in a number of fatalities the test was ordered when a patient’s condition deteriorated and it was clear that he or she was quite sick.
Although it would certainly be nice to have a simple test that could indicate on presentation which overdose patient was at higher risk of dying, I’m afraid that such an evaluation will continue to require complex clinical judgment, based on history, physical, vital signs, and all appropriate laboratory values. I very much doubt that lactate by itself will add much to the mix.
Dr. J Says:
Won’t elevated lactate be a potential marker for elevated mortality in just about any condition coming through the emergency room door?
I guess I understand the potential use in the same way we use lactate to look for ‘occult shock’ aka hypoperfusion with normal BP in patients with sepsis, we would be looking for the well looking OD that is going to go south. If we order a lactate on everyone with an overdose of anything what will we do with the information? If there is potential to offer early discharge for normal lactate patients, or early intensive therapy to elevated lactate patients I am all for it, but if lactate is just another non-actionable data point I question the utility.
August 26th, 2010
Leon Says:
Dr. J:
I agree completely. To make measuring lactate routinely in overdose patients worthwhile it would have to change management in a positive way. Frankly, I don’t see a chance of this ever coming to pass. Certainly if we check lactates on the sick-looking tox patients, many levels will be high — but we will have already known that they were sick. Also, since the incidence of death in tox cases is extremely low — well under 1% — any study designed to demonstrate that lactate could predict fatality would have to analyze an impossibly and absurdly large number of patients.
August 26th, 2010