cone
Ciriani
Arterial lactate levels in an emergency department
are associated with mortality: a prospective
> Adana mata s
Dublshed onine on. To iow
piece wth jour ole
thtpdedoo 10.1136)
‘nee 2013703541)
Emerge Medicine Resesch
Grup Einurgh EMERGE
Deparment of Erergerey
Medn, Roja iran of
Fein OHS Lain, le
France Gescet, Edneurh
ue
“cial Care Deparment,
Sth’ Hospital (NS
{atin ington, Uk
Belem Tr Cini
seach Fadi, ese
Sever Hosp, Unies of
dng. Enough UK
Correspondence to
Dr Deepa Data
Emergency Medic Resear
GtoupEaburgh EMERGE,
Depart of Emergency
ecko, Roy may of
Ednusgh (HS Lotion, ie
Fence Cresent Erbugh,
Seatnd 6 454 Uk
eepatcadatanhs ret
een 28 December 2013,
Reseed 27 Novber 2014
acpi 28 Nebr 2014
Pablhed nine Fast,
18 anuay 2015,
CB) Linked
> opal 0.1136
‘reed 2014204305,
hop og/0.1136)
termed 2015201764
® CrossMark
a ct: bata 0, Wake G
Geo Aleta og Med
201532673-67,
observational cohort study
Deepankar Datta, Craig Walker,'"? Alasdair James Gray,’ Catriona Graham?
ABSTRACT
Objectives Lactate measurements ae routinely cared
‘ut in emergency departments and ae associated wth
increased meaty in septic patients. However, no
definitive esearch as been cared out ito iether
lactate measurements canbe used asa prognostic marker
in a ciniclly unwell population in the emergency
department.
Methods We cared oxt a prospective observational
cohort study in consecutive patients whose atl lactate
concentration was measured in the emergency department
ofa tertiary refeal hospital assessing 110 000 patients
per year between 11th May and 1th August 2011.
‘Te main outcome measure was 30-day moray
Results There were 120 deaths (16.1%) a 30 das
postattendance in our cohort of 747 patents. Multivariate
logistic regression revealed loner latte levels were
associated with 30-day surival: ORs for 30-day death
‘compared with lactate 24 were 0.125 (95% CI 0.068 to
(0.228) for lactate <2 ane 0,273 (95% C10.140 100.533)
{or lactate 2—<4, Kaplan-Meier analysis showed a sunival
ference when dividing lactate concenttations int srata
(0.0001), This survival dfeence was maintained when
septic diagnoses were taken into acount.
Conclusions A single arterial lactate measurement on
presentation tothe emergency department predicts 30-day
‘mortality independent of other measures of ines severity.
BACKGROUND
Elevated serum lactate is recognised as a marker of
tissue hypoperfusion and organ dysfunction." ?
Lactate is produced and cleared as part of normal
homeostasis.” A serum arterial lactate of less than
2 mmolL is considered normal.* However, in states
where there is a mismatch of oxygen supply and
demand, cellular lactate is produced through anaer-
‘bic pachways, It has been shown that in critically
ill patients, lactate is produced by organs chat do
not normally produce lactate as well as in areas of
tissue inflammation.” Additionally, elevated lactate
levels may be present before clinical signs of shock
are present (‘occult shock’)?
Serum lactate is used as a biomarker ro identity
patients with tisue hypoperfusion” and is measured
routinely in critically ill pariens presenting to the
emergency department (ED). Following the work of
Rivers et al and the Surviving Sepsis Campaign, the
‘measurement of lactate is integral to the early asses-
‘ment and resuscitation of septic patients *
Moreover, arerial lactate levels >4 mmol/L. have
been shown to have a higher 28-day mortality inde~
pendent of hypotension.” Hyperacraraemia is also
associated with increased mortality in patients with
‘What is already known on this subject?
Lactate measurements are routinely cared out in
emergency departments and are associated with
increased mortality in septic patients. However, no
definitive research has been carried out into
whether lactate measurements can be used as a
prognostic marker in a clnically unwell population
in the emergency department
What this study adds?
‘single arterial lactate in clinically unwel patients
‘on presentation tothe emergency department i 2
predictor of 30-day mortality independent of other
measures of ness severity.
SST elevation myocardial infarction," pulmonary
emboli! stroke! and orthopaedic trauma."
Additionally, there is increasing evidence that lactate
clearance is an important resuscitation end point in
sepsic and trauma patients and associated with
‘improved clinical ontcomes.'*"!*
Lactate measurement has given physicians in the
ED a quantitative marker of abnormal physiology
to support risk stratification and as a treatment end
point in sepsis and trauma. However, inital lactate
levels in patients presenting to EDs while fre~
quently performed have not been rigorously inves-
tigated for use as a prognostic factor outside of the
above population groups,
The aim of this study is v0 investigate whether a
single arterial lactate measurement can be used as
prognostic marker in a clinically unwell population
presenting tothe ED.
‘METHODS
Study design and setting
‘We performed a prospective observational cohort
study of consecutive patients presenting to the ED
‘who had an arterial lactate sample obtained during
initial clinical assessment in a single UK NHS ED
(110 000 adult atendances annually) for a planned
3:month period between Ith May and 11th
August 2011, No intervention or change to clinical
‘management was institated.
Participants and recruitment
Patients were included if during ED clinical assess-
ment they had a serum arterial lactate performed as
part of their routine care as judged by clinician dis-
cretion. Patients were excluded if they had
BMJ
Data, eral ei Med 2018 3267S-A77 510.1 36lenered
RSH evTable 1 Categorical anges for selected physiological data
‘erable Below normal Normal Above normal
rt ate Ui) <0 50-99 29
Spt 3 ro Ho) a0 100-199 719
8 Une) s #20 320
Bicarbonate mat) a nae >
Create malt) < fon0 210
previously been included in the study. All samples were taken
‘within 4 h of presentation to the ED.
Data sources, variables and missing data
Al samples ‘were analysed in the ED ABG analyses, an
Instrumentation Laboratory Gem Premier 3000 analyser.” The
ABG analyser was set upto require unique patent identifier for
cach sample enabling optimal patient capture. Unidentifable,
venous and serial samples were excluded. Wisables obtained
from the analyser incinded lactate, PO,, PCOs, H+, bicarbonate
and standard base excess.
Patients with valid arterial samples were recorded and
followed-up through the ED and hospital parent administration
system (TakCare, InterSystems Corporation, Cambridge,
Massachusetts, USA). Data variables collected included: patient
demographics, past medical hor, vial signs on ED presentation,
Inboratory blood samples during ED asessment, presumed diag
nosis on ED discharge, hospital discharge diagnosis, source of
Table 2 Patent characteristics by 30-day survival
patient and patient disposition from the ED. Data were also cok
lected on 30-day mortality and 30-day critical care admision
Critical care was defined as admission to level 2(high-dependency
unit) or level 3 (icensive care) facilites
All missing data were recorded as an empry value, All data
thar were measured above ot below an extreme of reference
range were recorded at chat extreme range,
Statistical analysis,
Comparisons were made between subjects alive at 30 days and
those dead at 30 days. The variables ro be analysed were limited
by the authors to parameters that were noted to be the most
clinically significant and easily available to prevent type I erzors
bby multiple comparisons. Physiological variables were cate
gorised into low, normal and high categories; due to. small
group sizes heart rate, systolic BP and RR have only been con-
sidered with earegories (table 1).
Categorical variables where more than two categoties were
present were analysed using a 3 testy p values from Fishers
‘exact test were presented where appropriate due %0 small
‘counts. For variables with two categories, results are presented
8 ORs with 95%6 Cls
[A mutivariate logistic regression model was created to detet=
‘mine if arterial Lactate concentration is associated with 30-day
‘mortality when adjusted for any other factors. Variables with a
univariate p value of <0.1 were used in the development of the
regression model, which was then further refined by exchiding
variables which did not significantly improve the model. The
final analysis used age category, shock (defined as reported
‘lve ot 30 dys Dead at 30 days
nce) a0) Totel__ORof death (95%) pValue
Al pats 27 39) 0 (161) 7
Sender Mule 319183) 1167) 3B 110075016) on
Ae group attendance