venous and arterial blood gas analysis in the ed: what we know and what we don't
DESCRIPTION
This presentation delivered at the International Conference on Emergency Medicine in Dublin summarises agreement between venous and arterial blood gas parameters and utility of venous blood gas analysis in emergency department clinical practice. It also highlights important gaps in our knowledge on this topic.TRANSCRIPT
VENOUS AND ARTERIAL BLOOD GAS ANALYSIS IN EMERGENCY DEPARTMENTS:
WHAT WE KNOW AND WHAT WE DON’T!
Anne-Maree KellyProfessor and DirectorJoseph Epstein Centre for Emergency Medicine Research @Western Health
Permissions
This presentation can be used in part or whole for educational purposes on the condition that the following appears on each slide used:
‘Re-produced with permission of Professor Anne-Maree Kelly, Joseph Epstein Centre for Emergency Medicine Research, Melbourne, Australia’
@kellyam_jec
Conflicts of interest
I received financial support for travel and accommodation from Radiometer Pty Ltd to present a similar presentation at 4th International Symposium on Blood Gas and Critical Care in France in 2008.
I am undertaking some research with A/Prof Rees into calculated values which may be commercialised. I have no pecuniary interest in this program.
I have not received industry funding for any of my blood gas research projects.
Objectives
After this presentation, participants will:
Understand the agreement performance of variables on arterial and venous blood gas analysis, in particular
pH pCO2
Bicarbonate Base excess
Be aware of the unanswered questions Be aware of new approaches being taken to improve accuracy of
prediction of arterial values from venous blood gas samples
Caveats
Discussion will be limited to comparisons between arterial and peripheral venous samples
Data is up-to-date as of publications to May 2012 Includes some of data only ‘published’ as abstract in
2012
Blood gases in emergency medicine
Establishing acid-base status Mainly pH; but also bicarbonate
Measuring respiratory function/ ventilation Mainly pCO2; but also pH
‘Quick check’ potassium, haematocrit, some electrolytes Not addressed in this presentation
Why venous rather than arterial?
Less pain for patients Fewer complications, especially vascular and
infection Fewer needle-stick injuries Easier blood draw Minimal training requirement
Setting the context
JANE
26 year old, insulin dependent diabetic
2 days of vomiting and diarrhoea.
Pulse 120 bpm, BP 100/-, BSL ‘Hi’
TRAN
74 year old COAD Acute respiratory distress.
Pulse 110, BP 140/-, SpO2
(air) 88%
The clinical questions
Can we Exclude / diagnose Monitor progress of Base therapeutic decisions for (eg use and settings of
NIV)
Metabolic acidosis or acute respiratory failure using venous blood gas analysis rather than arterial?
Statistical considerations
Outcome of interest is how closely venous and arterial values agree, not how well they correlate
Weighted mean difference gives an estimate of the accuracy between the methods
95% limits of agreement give information about precision
Arterial value
Venous value
95% LoA
Clinical considerations
There is limited data about the tolerance clinicians have with respect to agreement between arterial and venous values of blood gas parameters
Depending on this tolerance, the degree of agreement may be acceptable or unacceptable There is known variation between clinicians re this
Issues with the evidence
Patient cohorts highly varied
Patient groups of interest are those at high risk of acidosis or hypercarbia Reporting does not always report this detail Data may to be dominated by patients with normal pH,
pCO2 and blood pressure Need for research focussed on high risk patient groups
pH
13 studies Range from 44 to 346 patients
Various conditions DKA (3), COAD (4), trauma (1)
2009 patients
Weighted mean difference of 0.033 pH units
95% limits of agreement (7 studies) generally within +/- 0.1 pH units
pH in illness subgroups
DKA
3 studies (265 patients) Weighted mean
difference = 0.02 pH units
95% limits of agreement = -0.009 to 0.02 pH units (1 study)
COAD
5 studies (643 patients) Weighted mean
difference= 0.034 pH units
95% limits of agreement generally +/- 0.1 (3 studies)
pH- Other
One ICU-based study suggests that as hypotension increases, AV pH agreement deteriorates Very small patient numbers Finding not yet validated
What we know & evidence gaps
We know: Generally close AV agreement in both respiratory and
metabolic disease
Evidence gaps: AV agreement in various levels and types of shock AV difference in toxicology scenarios (1 small study in
TCA OD only) AV difference in mixed acid-base disease
Bicarbonate
8 studies
1211 patients
Various conditions (COAD 2)
Weighted mean difference = -1.3mmol/l
95% limits of agreement : up to +/- 5mmol/l (3 studies)
Bicarbonate in illness subgroups
DKA
1 study (21 patients) Weighted mean
difference = -1.88 mmol/l
95% limits of agreement = -2.8 to 0.9 mmol/l
COAD
2 studies (643 patients) Weighted mean
difference= -1.34 mmol/l 95% limits of agreement:
none reported
What we know and evidence gaps
We know: Limited data suggests good agreement
Very little data re limits of agreement ? +/- 5mmol/L
Evidence gaps: AV agreement in specific disease states AV agreement in various levels and types of shock AV difference in toxicology scenarios AV difference in mixed acid-base disease
pCO2
8 studies
965 patients
Various conditions (COAD 4)
Weighted mean difference = 6.2 mmHg
95% limits of agreement: up to -17.4 to +23.9 mmHg 5/7 studies reporting LoA report LoA band >20mmHg
pCO2 in COAD
4 studies
452 patients
Weighted man difference = 7.26 mmHg
95% limits of agreement: up to -14 to +26 All 3 studies that reported LoA report LoA band
>20mmHg
Venous pCO2: A screening test for hypercarbia?
Author, year No. Screening cut-off
Sens. Spec. NPV %ABG avoided
Kelly, 2002 196 45 100 57 100 43
Kelly, 2005 107 45 100 47 100 29
Ak, 2006 132 45 100 * 100 33
McCanny, 2011
94 45 100 34 100 23
POOLED DATA
529
45 100 (95% CI 97-100)
53(95% CI 57-58)
100(95% CI 97-100)
35%(95% CI 32-41)
Data limited to studies in cohorts with respiratory disease
Using venous pH and CO2 to track progress?
Preliminary data presented at this meeting as a poster
41 comparisons in 29 patients Arteriovenous difference for change in pH =0.004 (95%
LoA -0.09 to 0.1) Arteriovenous difference for change in pCO2 = 0.55mmHg
(95% LoA -16.6 to 17.6mmHg)
What we know & evidence gaps
We know: AV agreement is NOT good enough for clinical inter-
changeability Wide limits of agreement
Venous pCO2 has potential as a screening test for hypercarbia Excellent NPV AV agreement in change in pCO2 is NOT good enough for clinical
inter-changeability (pilot data only) Wide limits of agreement
Evidence gaps: Whether trend in venous pCO2 and pH can safely drive a care pathway
for COAD Subject of current international research project
Base excess
Two studies only In a sample of 103 patients (various conditions), they
report: mean difference of 0.089mmol/L 95% limits of agreement -0.974 to +0.552 mmol/L
In 326 trauma patients mean difference -0.3 BE units 95% limits of agreement -4.4 to +3.9 BE units 20% did not fall within pre-defined clinical equivalence threshold
Current view: LOA too wide. If accuracy needed in critically ill, need ABG
Clinical application
JANE
DKA AV agreement is
acceptable; at least in non-shocked patients
Can use venous pH to diagnose/ monitor
TRAN
Acute respiratory distress
pH agreement good but pCO2 has considerable imprecision
Can use venous pCO2 as a screening test for hypercarbia
Another approach
Team from Center for Model Based Medical Decision Support Systems, Dept of Health Science and Technology, Aalborg University, Denmark (A/Prof Steven Rees)
Developed venous to arterial conversion method using venous blood gas variables and pulse oximetry
Designed to be incorporated into blood gas analysers
The model
The method calculates arterial values using mathematical models
Assumes: Constant value of the
respiratory quotient of 0.82 Change in base excess from
arterial to venous blood is 0 mmol/l
Rees SE, Toftegaard M, Andreassen S. A method for calculation of arterial acid–base and blood gas status from measurements in the peripheral venous blood. Comp Methods Programs Biomed. 2006, Vol 81, 18-25.
Validations
Respiratory patients N=40 (55% acute
admissions) Arterial-calculated pH
difference = -0.001pH units (95% LoA -0.026 to +0.026)
Arterial-calculated pCO2 difference = -0.68mmHg (95% LoA -4.81 to +3.45 mmHg)
Respiratory/ ICU N=103 Arterial-calculated pH
difference = -0.002pH units (95% LoA -0.029 to +0.025)
Arterial-calculated pCO2 difference = 0.3mmHg (95% LoA -3.58 to +4.18 mmHg)
Toftegaard et al. Emergency Medicine Journal. 2009;26:268-72
Rees et al. Eur Respir J. 2009;33:1141-7.
Validations
Emergency dept patients
N=148 patients (47 clinical need for ABG, 101 without)
pH can be calculated to within 0.02 pH units (95% LoA)
pCO2 can be calculated to within 4mmHg (0.5kPa)
Tygesen et al. Eur J Emerg Med. 2011 Nov 11. [Epub ahead of print]
Monitoring over time: Example
Red=measured arterialBlack dots =calculated arterialBlue dashes=measured venous
pH pCO2
Courtesy of SE Rees (unpublished)
Comments
Hard to know how many patients were acidotic or hypercarbic
No validation in patients undergoing respiratory support e.g. NIV
Model undergoing commercialisation Add on licence to blood gas machine No app planned at this stage (personal communication)
Take home messages
pH and bicarbonate probably close enough agreement for clinical
purposes in DKA, acute respiratory failure, isolated metabolic acidosis
More work needed in toxicology, shock, mixed disease
Take home messages
pCO2 NOT enough agreement for clinical purposes, either as
one-off or to monitor change Data suggests venous pCO2 is useful as a screening test
Base excess Probably not enough agreement for clinical purposes
Take home messages
Mathematical modelling approaches might be more accurate especially for pCO2
For broad applicability an app/ similar would be more feasible than integration into blood gas machines
More work needed to prove accuracy and precision in high risk groups
Questions?
Questions?
Questions?
@kellyam_jec