ARTICLE IN PRESS

Respiratory Medicine (2006) 100, 682–686

KEYWORDCOPD;Acute exaCapillary b

0954-6111/$ - sdoi:10.1016/j.r

�CorrespondE-mail addr

Capillary blood gases in acute exacerbations ofCOPD

Ross Murphy�, Sanjeet Thethy, Simon Raby, Jean Beckley,John Terrace, Christine Fiddler, Michelle Craig, Colin Robertson

Department of Accident and Emergency Medicine, The Royal Infirmary of Edinburgh,Old Dalkeith Road, Little France, Edinburgh EH16 4SU, UK

Received 15 June 2005; accepted 21 July 2005

S

cerbation;lood gases

ee front matter & 2005med.2005.08.007

ing author. Tel.: +44 771ess: rossmurphy@docto

SummaryObjectives: To assess the correlation and agreement between measurements ofPO2, PCO2, H+ and HCO3

� in arterial and capillary blood in patients with acuteexacerbations of COPD. To assess the repeatability of capillary measurements.Design: Method comparison study.Setting: Accident and emergency department in a university teaching hospital.Main outcome measures: Measurements of PO2, PCO2, H

+ and HCO3� in one arterial

and two capillary samples taken from consecutive patients with acute exacerbationsof COPD.Results: The agreement between measurements of PCO2, H

+ and HCO3� in arterial

and capillary blood was good with mean differences of 0.087 kPa, 1.044 nmol/l and0.513mmol/l, respectively. The corresponding 95% limits of agreement were narrow.The agreement between measurements of PO2 was poor with a mean difference of1.256 kPa and wide 95% limits of agreement. There was good repeatability betweencapillary samples with mean differences of 0.094 kPa, 0.674 nmol/l and 0.028mmol/l for measurements of PCO2, H+ and HCO3 respectively and narrow coefficients ofrepeatability.Conclusions: Capillary blood gas measurements provide an accurate assessment ofPCO2, H+ and HCO3

� and can be used to reliably measure the ventilatory status ofpatients. Combined with continuous pulse oximetry they can be used as analternative to arterial blood gas measurements in patients with acute exacerbationsof COPD.& 2005 Published by Elsevier Ltd.

Published by Elsevier Ltd.

2 761172.rs.org.uk (R. Murphy).

ARTICLE IN PRESS

Capillary blood gases in acute exacerbations of COPD 683

Introduction

Acute exacerbations of COPD account for a largepercentage of presentations to Emergency Depart-ments in the UK each year. A crucial part of everypatient’s assessment is measurement of arterialblood gases.1 This is a painful and uncomfortableprocedure with potentially dangerous complica-tions.2

Capillary blood gases have been used for manyyears in children as a less distressing and saferalternative and some studies suggest that theyagree well with arterial values.3 Varying levels ofagreement between arterial and capillary samplesare found in adult patients with different pathol-ogies.4 It is felt that this agreement can beimproved if the capillary sampling site is first‘‘arterialised’’ by applying vasodilating pastes orwarming it.4

It has not been determined as to whether, inpatients with acute exacerbations of COPD, capil-lary blood gases are a viable alternative. Patientswith COPD most often need blood gas analysiswithin an Emergency Department setting.

In this study we assessed the correlation andlevel of agreement between measurements of PO2,PCO2, H+ and calculated HCO3

� in arterial andcapillary blood in patients with acute exacerba-tions of COPD.

0 8 10 12 14

0

2

4

6

8

10

12

14

Arterial (kPa)

Cap

illar

y (k

Pa)

642

Figure 1 PCO2 measurements with line of equality.

Methods

The study was approved by the Lothian ResearchEthics Committee and was performed in theAccident and Emergency department at the RoyalInfirmary of Edinburgh between September 2002and February 2003.

Consecutive patients presenting with a recentincrease in cough, wheeze or shortness of breathand a prior diagnosis of COPD, made by a consultantphysician were eligible for inclusion. Patients wereexcluded if there was an obvious alternative causefor their respiratory symptoms.

Enrolment was performed whenever the studyauthors, who represent one of three teams provid-ing a 24 h service, were in the department.

Informed written consent was obtained priorto entry. Some patients were drowsy or confuseddue to their illness and in these cases informedwritten assent was obtained from accompanyingrelatives.

It was aimed to recruit over 50 patients as this isthe minimum sample size recommended for amethod comparison study.5

On entry, patients had a vasodilating nicotinatepaste applied to their earlobe. After 10min thepaste was removed and the inferolateral part of theearlobe pricked with a sterile needle. The sur-rounding tissue was squeezed and the bloodcollected into two capillary tubes. Two sampleswere taken in order to assess the repeatability ofcapillary sampling. Immediately afterwards anarterial sample was taken from the radial arteryin order to assess the agreement between thecapillary and arterial samples. The samples wereaspirated in the same order into the department’sblood gas analyser (GEM premier 3000, Instrumen-tation Laboratory) and the blood gas measurementsfor each sample recorded.

Each patient was treated appropriately for theircondition with oxygen, nebulised bronchodilatorsand other treatments deemed appropriate by asecond attending doctor.

Results

Sixty patients were eligible for entry into thestudy. Three refused consent and two were inad-vertently not considered leaving a study sample of55. In 15 of these cases assent was given byrelatives and in 40 consent was given by thepatient. In five patients a second capillary samplewas not obtained.

The results are illustrated in Figs. 1–4 on whicheach first capillary measurement is plotted againsteach arterial measurement. They also show the lineof equality on which all points would lie if therewere exact agreement. A similar evaluation wasperformed between the two sets of capillarymeasurements.

The results were further analysed using methodsdescribed by Bland and Altman to assess theagreement between two methods of clinical

ARTICLE IN PRESS

151050

1

0

-1

Average between each arterial andcapillary measurement (kPa)

Diff

eren

ce b

etw

een

arte

rial a

ndca

pilla

ry m

easu

rem

ents

(kP

a) Mean + 2SD

Mean

Mean - 2SD

Figure 5 Mean difference and limits of agreements forPCO2.

6050403020100

5

0

-5

Average between each arterial andcapillary measurement (nmol/l)

Diff

eren

ce b

etw

een

arte

rial a

ndca

pilla

ry m

easu

rem

ents

(nm

ol/l) Mean + 2SD

Mean

Mean - 2SD

Figure 6 Mean difference and limits of agreements forH+.

706050403020100

70

60

50

40

30

20

10

0

Arterial (nmol/l)

Cap

illar

y (n

mol

/l)

Figure 2 H+ measurements with line of equality.

6050403020100

60

50

40

30

20

10

0

Arterial (mmol/l)

Cap

illar

y (m

mol

/l)

Figure 3 HCO3� measurements with line of equality.

403020100

40

30

20

10

0

Arterial (kPa)

Cap

illar

y (k

Pa)

Figure 4 PO2 measurements with line of equality.

50403020100

5

4

3

2

1

0

-1

-2

-3

-4

-5

Average between each arterial andcapillary measurement (mmol/l)

Diff

eren

ce b

etw

een

arte

rial a

ndca

pilla

ry m

easu

rem

ents

(m

mol

/l)

Mean + 2SD

Mean

Mean - 2SD

Figure 7 Mean difference and limits of agreements forHCO3

�.

R. Murphy et al.684

measurement.6 The results of the analysis betweenthe arterial samples and the first set of capillarysamples are illustrated in Figs. 5–8. For measure-ments of PCO2, H

+, HCO3� and PO2 95% of capillary

values would expect to fall within 2 standard

deviations above and below the mean differenceswhen compared with arterial measurements. Theseare the 95% limits of agreement.

The results of all the above are summarised inTable 1.

The results of the analysis of repeatabilitybetween the 2 sets of capillary measurements aresummarised in Table 2.

ARTICLE IN PRESS

Capillary blood gases in acute exacerbations of COPD 685

Discussion

This study is the first to show that for measure-ments of PCO2, H

+ and HCO3� arterialised capillary

blood samples agree well with arterial bloodsamples in patients with acute exacerbations ofCOPD. For PO2 the agreement is insufficient toallow capillary sampling to replace standard ana-lysis in clinical use.

These results are consistent with previous re-ports that have suggested a use for capillary bloodgases as a means of estimating the ventilatorystatus of patients but not the degree of oxygena-tion.4,7

Table 1 Agreement between arterial and first capillary

Measurement Correlationcoefficient (r)

Mean differe

PCO2 (kPa) 0.984 0.087H+ (nmol/l) 0.961 1.044HCO3

� (mmol/l) 0.977 �0.513PO2 (kPa) 0.830 1.256

Table 2 Agreement between two capillary samples.

Measurement Correlationcoefficient (r)

Mean differe

PCO2 (kPa) 0.993 0.094H+ (nmol/l) 0.964 0.674HCO3

� (mmol/l) 0.992 0.028PO2 (kPa) 0.967 0.062

3020100

25

20

15

10

5

0

-5

-10

-15

-20

-25

Average between each arterial andcapillary measurement (kPa)

Diff

eren

ce b

etw

een

arte

rial a

ndca

pilla

ry m

easu

rem

ents

(kP

a)

Mean + 2SD

Mean

Mean - 2SD

Figure 8 Mean difference and limits of agreements forPO2.

Although the study authors feel that the level ofagreement shown is good it is possible that inreality the agreement could be even better.

A vasodilating paste was used at the capillarysampling site to promote conformity between thearterial sample and the capillary sample and makethe sample easier to obtain. Other studies haveused methods such as heating the site with otherpastes, towels, lamps and water.8 The nicotinatepaste used in this study is cheap, easily obtainedand applied and fast-acting enough to be used indaily clinical practice.

It has been stated that the capillary sampling siteshould not be squeezed as this may result incontamination with venous blood, lymph and fattytissue.9 We found that unless the site was squeezedinsufficient blood was obtained for analysis.

It has also been recommended that metal filingsbe inserted into each capillary tube to aid samplemixing with a magnet.9 This was impractical anddifficult to reproduce. Care was taken with allsamples to ensure a continuous line of blood andthe sample was taken as quickly as possible to limitcontamination with room air.

Our results therefore represent capillary samplesthat are unmixed, arterialised with a simple pasteand taken quickly from a site that has beensqueezed. They can easily be performed in a busyEmergency Department.

As the capillary samples were taken and pro-cessed before the arterial sample it is possible thatthis increased discrepancy between measurements.

sample.

nce Standard deviation(SD)

95% limits ofagreement(Mean72SD)

0.370 0.827 to �0.6531.612 4.268 to �2.1801.364 2.215 to �3.2413.371 7.998 to �5.486

nce Standard Deviation(SD)

Coefficient ofrepeatability(Mean72SD)

0.228 0.550 to �0.3621.494 3.662 to �2.3140.821 1.67 to �1.6141.088 2.238 to �2.114

ARTICLE IN PRESS

R. Murphy et al.686

When a patient first had capillary sampling theymay have hyperventilated due to the slightdiscomfort resulting in an increase in PO2 and adecrease in PCO2 before the arterial sample wastaken. It is also possible that the arterial sampledegraded while waiting for the capillary samples tobe processed further increasing any discrepancy.

An important part of the assessment of any newmeasurement method is the assessment of itsrepeatability.6 In this study we took and processedtwo capillary samples consecutively. These wouldbe subject to the same sources of error as above.We found that the coefficient of repeatability forthe capillary samples was small indicating goodrepeatability. We did not assess the repeatability ofthe arterial samples.

The first step in the management of patients withacute exacerbations of COPD is ensuring adequateand controlled oxygenation. It is recommendedthat patients receive enough oxygen to consistentlyachieve a target oxygen saturation of about 90–92%or lower in some cases.10 Continuous pulse oxime-try is preferable to blood gas analysis for this inmost instances. It is non-invasive, painless and canbe performed by any trained personnel. Providedthere is a good tracing it will give an acceptableestimate of oxygenation, will rapidly detect hypox-ia and provides continuous information on oxygena-tion rather than a single estimate which will besubject to transient changes in a patients ventila-tory pattern.11

Subsequent steps involve establishing the venti-latory status by measuring PCO2, H

+ and HCO3� and

using these to establish prognosis and decide if non-invasive or invasive ventilation is indicated. Oftenrepeated measurements are needed. This studyshows that this can be done using capillary bloodgases. They are less invasive, less painful and moreeasily performed by trained personnel than arterialsamples.

Combined with continuous pulse oximetry, capil-lary blood gases can be used as an alternative toarterial blood gases in the initial assessment andsubsequent management of all patients with acuteexacerbations of COPD.

Contributors

Ross Murphy and Colin Robertson designed thestudy, recruited patients, analysed the results andwrote the paper. Sanjeet Thethy recruited pa-tients, analysed the results and wrote the paper. Allthe other authors recruited patients and wrote thepaper. Ross Murphy acts as guarantor.

Acknowlegements

The study was funded by a grant from the BritishAssociation for Accident and Emergency Medicine.

References

1. British Thoracic Society. BTS guidelines for the managementof COPD. Thorax 1997;52(Suppl 5):S1–S28.

2. Flenley DC. Arterial puncture. In: Procedures in practice,2nd ed. London: British Medical Journal, 1988. p.3–5.

3. Escalante-Kanashiro R, Tantalean-Da-Fieno J. Capillaryblood gases in a paediatric intensive care unit. Crit CareMed 2000;28(1):224–6.

4. Pitkin AD, Roberts CM, Wedzicha JA. Arterialised earlobeblood gas analysis: an underused technique. Thorax1994;49:364–6.

5. Altman DG. Some common problems in medical research. In:Altman DG, editor. Practical statistics for medical research.1st ed. London/Boca Raton, FL: Chapman & Hall/CRC; 1991.p. 396–403.

6. Bland JM, Altman DG. Statistical methods for assessingagreement between two methods of clinical measurement.The Lancet 1986;i:307–10.

7. Eaton T, Rudkin S, Garrett JE. The clinical utility ofarterialised earlobe capillary blood in the assessment ofpatients for long-term oxygen therapy. Respir Med 2001;95(8):655–60.

8. Christoforides C, Miller JM. Clinical use and limitations ofarterialised capillary blood for PO2 determination. Am RevRespir Dis 1968;98:653–7.

9. AARC clinical practice guideline. Respir Care 1994;39(12):1180–3.

10. Murphy R, Mackway-Jones K, Sammy I, et al. Emergencyoxygen therapy for the breathless patient. Guidelinesprepared by the North-West Oxygen Group. EmergencyMedicine Journal 2001;18(6):421–3.

11. Hanning CD, Alexander-Williams JM. Pulse oximetry:a practical review. BMJ 1995;311:367–70.