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

- Rogue Medic

Nontraumatic out-of-hospital hypotension predicts inhospital mortality

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

An interesting examination of something that we take for granted. Does any instance of hypotension increase the risk of death for patients with life-threatening or potentially life-threatening conditions? Hypotension is categorized as SBP (Systolic Blood Pressure) less than 100 mm Hg, rather than SBP <90.

They assessed patients with respiratory distress, syncope, chest pain, dizziness, altered mental status, anxiety, thirst, weakness, fatigue, or the sensation of impending doom.

Due to the difference in age of the groups, the non-hypotensive group was abbreviated (truncated) to match the significantly older (P<.0001, unpaired t test) hypotensive group within one standard deviation. This cut the non-hypotensive group from 2,733 to 1,362 – eliminating just over half of the group. This should do a good job of controlling for the age difference. There were two locations for the study, but the significant difference in ages was only observed at one location.

Nonexposure patients were thus priority I or II transported patients, aged 48 to 84 years, with systolic blood pressure always more than 100 mm Hg and with 1 or more of 10 predefined symptom documented.[1]

That misrepresents the nonexposure (not hypotensive) patients. These patients did not have continual measurement of their blood pressures. A minimum of only one set of vital signs was required.

I do try to take vital signs when there is a change in patient presentation, but I have noticed that not everyone behaves as I do.

A lack of documentation of <100 SBP is not the same as systolic blood pressure always more than 100 mm Hg. Here are some of the problems with assuming that all <100 SBPs were identified by a random assessment of at least one set of vital signs –

Were vital signs assessed with every change in presentation?

Were all changes in presentation observed?

Is a drop in SBP always going to be accompanied by a change in presentation?

Is hypotension always going to be accompanied by a change in presentation?

The answers are – No, No, No, and No.

Inhospital mortality was determined by first searching the Social Security Death Index.[1]

Is this a good method of differentiating between living people and dead people?

If the government thinks I am alive, that does not mean that I am alive. If the government thinks I am dead, that does not mean that I am dead, nor that I have a <100 SBP.

Secondary analysis measurements were the relationship between age and inhospital mortality in hypotensive patients, the relationship between the initial out-of-hospital systolic blood pressure and inhospital mortality, the inhospital mortality rate of transient versus sustained hypotension, and a calculation of the sensitivity and specificity of out-of-hospital hypotension for inhospital mortality.[1]

At one site, SBP from 91 to 99 appears to improve survival. This may be just the law of small numbers at work. Those with transient hypotension are probably going to have blood pressures that are close to the dividing line.

Venue 2 has similar results, except here there are larger increases in the death rates at 91 – 99 and <70.

The sustained low blood pressure appears to be the real danger, while transient low blood pressure does not appear to be significantly different from zero assessments of low blood pressures.

the mortality rate among these 255 patients was 33% (95% CI 27% to 39%).[1]

The effect of a sustained low blood pressure would have been a good study to follow this study.

Is it predictive?

Does it predict something that we can reverse?

Does sustained low blood pressure mean sustained in spite of treatment?

What kind of treatment(s) would be appropriate?

Patients with a systolic blood pressure equal to 100 mm Hg were not included in the analysis; at venue 1 this was 23 patients, and at venue 2 it was 252 patients.[1]

Those patients would have had one or more measurements of SBP = 100, but no measurements of SBP <100. 5.8% (regardless of group) at venue 1 and 3.5% of the nonexposure group at venue 2. The 252 patients would have added almost another half (47.3%) to the exposure group at venue 2.

This raises an important question.

Why is this one SBP so commonly represented as the lowest blood pressure in the vital signs?

How often will 100 be the lowest documented SBP in any group?

Is <100 only a little more than twice as likely as 100 for the lowest documented SBP?

Does documentation of <100 SBP require treatment under a different protocol/algorithm?

There were 3,128 patients at venue 1 and only 23 SBP measurements of exactly 100.

There were 7,679 patients at venue 2 and 252 SBP measurements of exactly 100.

Venue 2 has only 2 1/2 times as many patients, but 11 times as many SBP measurements of exactly 100. Were the exclusions for SBP exactly equal to 100 done before truncation and matching? We cannot tell from the information provided.


[1] Nontraumatic out-of-hospital hypotension predicts inhospital mortality.
Jones AE, Stiell IG, Nesbitt LP, Spaite DW, Hasan N, Watts BA, Kline JA.
Ann Emerg Med. 2004 Jan;43(1):106-13.
PMID: 14707949 [PubMed – indexed for MEDLINE]

Jones, A., Stiell, I., Nesbitt, L., Spaite, D., Hasan, N., Watts, B., & Kline, J. (2004). Nontraumatic out-of-hospital hypotension predicts inhospital mortality☆ Annals of Emergency Medicine, 43 (1), 106-113 DOI: 10.1016/j.annemergmed.2003.08.008