ABG Versus VBG in the Emergency Department

Authors: Rachel Kelly, MD (EM Resident Physician, Stony Brook University Hospital); Robert Nocito, MD (EM Attending Physician, Stony Brook University Hospital) // Reviewed by: Jessica Pelletier, DO (EM Education Fellow, Washington University in St. Louis); Marina Boushra, MD (EM-CCM, Cleveland Clinic Foundation); Brit Long, MD (@long_brit)

Case

Emergency Medical Services brings in a 62-year-old male with COPD in acute on chronic hypoxemic respiratory failure (usually on 3 L nasal cannula, now on non-rebreather at 15 L/min). As the only respiratory therapist in the ED has been paged and is starting BiPAP for this patient, an overhead call for two incoming trauma alerts from a multivehicle collision sounds. Because the RT responsible for drawing arterial blood gases is busy caring for these patients, ABGs will be delayed. How would an ABG even change the initial stabilization of any of these patients? What alternative testing can be done in these circumstances?

Introduction

Arterial blood gas (ABG) or venous blood gas (VBG) testing is used to assess the pH and systemic carbon dioxide tension in patients, and, therefore, provide a more complete picture of their acid-base status than an isolated basic metabolic panel (BMP) (1). ABGs have traditionally been considered the preferred method compared to VBGs, as they are thought to be more accurate and reflective of true oxygenation status than VBGs (1). However, ABGs have many drawbacks compared to VBGs. ABGs are less readily available, more time-consuming, and very painful for patients. Additionally, arterial blood draws are associated with more complications than venous blood draws, including arterial occlusion, hematoma, aneurysm formation, thrombosis/embolism development, and trauma or damage to adjacent nerve structures (1, 2). Recent studies suggest that clinicians can obtain similarly actionable results from VBGs as ABGs and using the “gold standard” ABG might not be as clinically relevant as it once was.

ABG and VBG Correlation

The correlation between venous and arterial blood gases is well-documented for standard differences (Table 1), and the data obtained from the VBG can be acted on as if it were an ABG (1, 3-6). Importantly, the comparison between an ABG and VBG is site-specific, as illustrated below (Table 1) (1, 3-6). In general, VBG pH is slightly more acidotic, and venous gases have a slightly higher PCO₂ but these differences are generally not clinically relevant except if the PCO₂ is significantly elevated (over 45 mm Hg), in which case ABG is necessary if the specific PCO₂ value is needed (7).  The partial pressure of oxygen in arterial gases cannot be extrapolated from either central or peripheral venous gases. Of note, HCO₃ in both ABGs and VBGs are calculated using the Henderson-Hasselbach equation and not directly measured; while the calculated values in an ABG or a VBG are often close to what is measured, an accurate measurement should be obtained from a concomitant basic metabolic panel (BMP).

*Peripheral obtained from IV line; central obtained from central venous line.

While ABGs and VBGs usually correlate well, there are a few situations where this correlation is disturbed. First, there can be a poor correlation between venous and arterial gases in states of shock, hemodynamic instability, or extremes of acid-base disturbances, due to poor cardiac output/circulatory failure impairing venous and arterial flow, with associated poor gas exchange as well as hypercapnia and acidemia at the tissue level (8). In terms of acid-base status, a VBG is still useful to screen for acid-base abnormalities, because the pH in both gases has been shown to correlate well across all values. However, there is better agreement between ABG and VBG results at normal values (i.e., when the pH of the VBG is normal, it is more likely to correlate to an ABG with a normal pH, compared to when the VBG pH is outside of normal values) (3). The venous PCO₂ on VBG may be increased threefold to fourfold relative to the arterial value due to alterations in circulation and perfusion. However, a screening VBG can still be obtained, and if within normal limits, the ABG results would likely be similar (3, 4, 8). An ABG can always be obtained after a screening VBG if clinical suspicion for an acid-base derangement or hypercarbia remains high. Notably, there is data that even in hypotensive patients a VBG is a reasonable alternative to an ABG, and mathematical regression equations can be used to derive data/values for pH, pCO₂, HCO₃, and lactate from the VBG if needed, though the clinical utility and practicality remains to be determined (2).

The other scenario where obtaining an ABG might be necessary is when serial VBGs are being obtained for patient monitoring and determining ongoing clinical course, especially in patients on prolonged mechanical ventilation (5, 9). In this situation, an ABG should be obtained periodically for correlation, though this is more relevant for the intensive care unit (ICU) setting than in routine ED care (5, 9).

Bottom Line: VBGs, whether obtained peripherally or centrally, tend to be slightly more acidotic and exhibit higher PCO₂ values than concomitant ABGs. VBGs have been found to correlate well with ABGs, but caution should be used in the extremes of hemodynamic instability.

Supplementing information from VBGs

Unlike pH, PCO₂, and HCO₃, there are no correlations or conversions that reliably determine oxygenation status from a VBG (1, 10). The pulse oximetry device is a noninvasive and continuous way of measuring peripheral arterial oxygenation saturation. This is why pulse oximetry has been used to supplement the VBG, as the combination of both modalities has been shown to determine oxygenation status with more accuracy (10).

The pulse oximeter has its drawbacks because it does not directly measure PaO₂ but serves only as a surrogate marker for oxygenation. When the pulse oximeter is compared to an ABG, it has been found that SpO₂ values of 90% and above tend to be 2-3% above the true arterial value. This correlation between ABG values worsens with lower SpO₂ readings (9-11). This makes VBG less useful for determining SpO₂ status in critical/unstable and septic patients, and those receiving supplemental oxygen (as it can hide clinically significant hypoxemia) (9-11). Additionally, in the setting of poor extremity perfusion – whether from vasoconstriction, peripheral vascular disease, or hemodynamic instability (especially hypotensive patients with SBP <80 mmHg) – the pulse oximeter can give falsely low values that do not correlate with PaO₂ levels (9-11)

A prospective cohort study looking at undifferentiated critically ill patients in both ICU and ED settings found that the combination of VBG and pulse oximetry provided adequate information compared to what the gold standard of an ABG could provide to make informed clinical decisions regarding oxygenation, ventilation, and acid-base status (7). Notably, the diagnosis of acute respiratory distress syndrome (ARDS) has classically relied on obtaining an ABG to calculate the PaO₂/FiO₂ ratio. It has been found that using pulse oximetry to obtain the SpO₂/FiO₂ ratio could help facilitate earlier ARDS recognition as pulse oximetry is more readily available and less time consuming to obtain compared with ABG, and this has been incorporated into the new proposed 2023 definition of ARDS (12). While this could be useful in resource-limited settings, it still requires additional studies and trials validating the SpO₂/FiO₂ ratio to satisfy the current PaO₂/FiO₂ ratio criterion for ARDS, as well as additional subsequent interventions based on the severity of ARDS (i.e., proning patients, ECMO) (12).

Bottom Line: For the majority of patients, clinicians can use VBG data supplemented with pulse oximetry to determine management. For unstable patients and those requiring escalating oxygen therapy, an ABG can be considered for more accurate titration of therapy but likely would not change patient disposition or ED management.

Clinical Utility of an ABG

Depending on ED and hospital institutional policies and location, ABGs are frequently obtained for trauma codes/activations and requested by ICU colleagues to determine patient disposition. Are there circumstances where ABGs are necessary, and why may consultants specifically request them?

Trauma:

In the trauma setting, metabolic acidosis often occurs secondary to hypoperfusion from hemorrhagic shock. It has been shown that pH, serum lactate, and base deficits are biomarkers of shock in trauma, correlating well with injury severity. These values have previously been used in prognostication, as it was previously thought that patients with pH <7.0 had almost universal mortality; however, recent studies have demonstrated that a pH < 6.6 is more likely to portend death (13, 14) . Trauma surgeons have traditionally relied on ABGs to assess acid-base status for overall prognostication of trauma injury severity and to screen for occult injury or malperfusion, as this might alter management and aggressive resuscitation. Trauma literature still supports obtaining routine screening and serial blood gases in all trauma patients (15, 16).

As previously noted, ABGs are not easily obtained (let alone in the critically ill trauma patient). Some studies have found that VBGs and ABGs do correlate well in trauma patients but are not clinically equivalent; these studies still call for obtaining routine ABGs in patients with major traumatic injuries until more research has been done comparing ABGs and VBGs in the setting of severe traumatic injury (17-19). A small prospective comparison study found that in the initial resuscitation of mechanically ventilated trauma patients, VBGs can provide clinically relevant information on pH, PCO₂, and base excess values, but for later resuscitation still state ABGs would be better for clinical management, as it is still viewed as the gold standard for ventilator adjustments for PCO₂ and pH, a conclusion that has been further supported with a more recent prospective observational study (the AGREE trial) (5, 20).

Despite its utility, data from an ABG is unlikely to change the initial course of trauma resuscitation, especially given the myriad of other ways for assessment and risk stratification of trauma severity available in the ED. There is well-documented evidence that low systolic blood pressure, a low Glasgow Coma Score (GCS), and a high Injury Severity Score (ISS) are known predictors of poor outcomes, and in combination with serum lactate provide adequate prognostic value. In contrast, ABG and pH alone are not reliable in determining the clinical futility of resuscitation after major trauma. (14, 21-23) Additionally, a study has shown that in trauma patients receiving extensive CT imaging from head to pelvis, the clinical management and disposition were not changed based on ABG and lactate levels when imaging did not reveal traumatic injuries, making routine ABG unnecessary in this population (24). In terms of traumatic brain injury, the ABG is not a good prognosticator of patient outcome nor does it correlate with GCS scores (25). Notably, in the initial analysis of blunt chest trauma (i.e. lung contusion, pneumothorax, and hemothorax) it has been found that an ABG demonstrating hypoxemia may be helpful for early identification of patients with lung injury (26).

Bottom Line: Trauma literature supports the use of ABGs due to a lack of studies showing that VBGs compare well to ABGs in trauma patients, but data from an ABG in isolation is unlikely to change management, aid disposition, or determine prognostication of continuing resuscitation in trauma patients.

Non-traumatic critically ill patients:

There are many formulas (i.e., acid-base disturbance/compensation, alveolar gas equation, A-a gradient) that are used in the management of critically ill patients which were developed using data from ABGs alone (27). This is especially relevant in ARDS patients or those in acute hypoxemic respiratory failure requiring escalating levels of oxygenation therapy, in whom these values are used to guide candidacy for and efficacy of therapy escalation. While VBG data cannot be used in calculations for these formulas, more research is showing that VBGs can be reliable in the assessment of these patients.

Recent studies have shown that obtaining a VBG to assess pH, bicarbonate, and PCO₂ is a reasonable alternative to ABG and provides values for calculations (i.e., acid-base disturbance/compensation) in the initial evaluation of an ED patient across all spectrums of disease assessed, and even in patients undergoing non-invasive ventilation (6, 28, 29). Obtaining routine ABGs in acutely dyspneic patients presenting to the ED does not alone provide much diagnostic or prognostic value; thus, VBG should be considered over ABG for the undifferentiated dyspneic patient (30).

In the common ED scenarios of chronic obstructive pulmonary disease (COPD) exacerbation and diabetic ketoacidosis (DKA), there is no solid evidence to support the practice of using ABGs over VBGs for management in these patients. This particularly holds true when it comes to initial ED management, and clinicians should feel reassured in using VBGs in combination with clinical assessment in making treatment, management, and disposition decisions in these patients (31). When evaluating for hypercarbia in COPD patients, a VBG should be used as a screening tool and ABG can likely be avoided since the sensitivity of VBG for pCO2 < 45 mmHg is as high as 100% (32, 33). In patients with DKA, a VBG is adequate to accurately assess the degree of metabolic acidosis, as ABGs have not been found to alter ED diagnosis, management, or disposition when compared with VBGs in this patient population (34, 35).

Bottom Line: While ABGs are useful for their use in prognostic calculations and utility in clinical decision-making for patients with ARDS, the literature supports the use of VBGs over ABGs in the treatment of most critically ill patients, particularly in the ED setting.

Pearls

– VBGs compare well to ABGs, especially in combination with supplemental pulse oximetry data, and the ED clinician should feel comfortable using them when indicated for medical and trauma patients.

– In extremes of hemodynamic instability consider obtaining a confirmatory ABG alongside your clinical acumen, though it likely will not change your ED management.

– Trauma and medical intensivists may prefer ABGs from a more traditional standpoint, but the ED clinician should feel reassured that obtaining an ABG generally will not alter management.

References/Further Reading

(1) Byrne AL, Bennett M, Chatterji R, Symons R, Pace NL, Thomas PS. Peripheral venous and arterial blood gas analysis in adults: are they comparable? A systematic review and meta-analysis. Respirology. 2014 Feb;19(2):168-175. doi: 10.1111/resp.12225. Epub 2014 Jan 3. PMID: 24383789.

(2) Prasad H, Vempalli N, Agrawal N, Ajun UN, Salam A, Subhra Datta S, Singhal A, Ranjan N, Shabeeba Sherin PP, Sundareshan G. Correlation and agreement between arterial and venous blood gas analysis in patients with hypotension-an emergency department-based cross-sectional study. Int J Emerg Med. 2023 Mar 10;16(1):18. doi: 10.1186/s12245-023-00486-0. PMID: 36899297; PMCID: PMC9999648.

(3) Bloom BM, Grundlingh J, Bestwick JP, Harris T. The role of venous blood gas in the emergency department: a systematic review and meta-analysis. Eur J Emerg Med. 2014 Apr;21(2):81-8. doi: 10.1097/MEJ.0b013e32836437cf. PMID: 23903783.

(4) Chong WH, Saha BK, Medarov BI. Comparing Central Venous Blood Gas to Arterial Blood Gas and Determining Its Utility in Critically Ill Patients: Narrative Review. Anesth Analg. 2021 Aug 1;133(2):374-378. doi: 10.1213/ANE.0000000000005501. PMID: 33780397.

(5) Malinoski DJ, Todd SR, Slone S, Mullins RJ, Schreiber MA. Correlation of central venous and arterial blood gas measurements in mechanically ventilated trauma patients. Arch Surg. 2005 Nov;140(11):1122-5. doi: 10.1001/archsurg.140.11.1122. PMID: 16342377.

(6) Malatesha G, Singh NK, Bharija A, Rehani B, Goel A. Comparison of arterial and venous pH, bicarbonate, PCO2 and PO2 in initial emergency department assessment. Emerg Med J. 2007 Aug;24(8):569-71. doi: 10.1136/emj.2007.046979. PMID: 17652681; PMCID: PMC2660085.

(7) Zeserson E, Goodgame B, Hess JD, Schultz K, Hoon C, Lamb K, Maheshwari V, Johnson S, Papas M, Reed J, Breyer M. Correlation of Venous Blood Gas and Pulse Oximetry With Arterial Blood Gas in the Undifferentiated Critically Ill Patient. J Intensive Care Med. 2018 Mar;33(3):176-181. doi: 10.1177/0885066616652597. Epub 2016 Jun 9. PMID: 27283009; PMCID: PMC5885755.

(8) Adrogué HJ, Rashad MN, Gorin AB, Yacoub J, Madias NE. Assessing acid-base status in circulatory failure. Differences between arterial and central venous blood. N Engl J Med. 1989 May 18;320(20):1312-6. doi: 10.1056/NEJM198905183202004. PMID: 2535633.

(9) Van de Louw A, Cracco C, Cerf C, Harf A, Duvaldestin P, Lemaire F, Brochard L. Accuracy of pulse oximetry in the intensive care unit. Intensive Care Med. 2001 Oct;27(10):1606-13. doi: 10.1007/s001340101064. PMID: 11685301.

(10) Jubran A. Pulse oximetry. Intensive Care Med. 2004 Nov;30(11):2017-20. doi: 10.1007/s00134-004-2399-x. Epub 2004 Jul 24. PMID: 15278272.

(11) Webb RK, Ralston AC, Runciman WB. Potential errors in pulse oximetry. II. Effects of changes in saturation and signal quality. Anaesthesia. 1991 Mar;46(3):207-12. doi: 10.1111/j.1365-2044.1991.tb09411.x. PMID: 2014899.

(12) Wick KD, Matthay MA, Ware LB. Pulse oximetry for the diagnosis and management of acute respiratory distress syndrome. Lancet Respir Med. 2022 Nov;10(11):1086-1098. doi: 10.1016/S2213-2600(22)00058-3. Epub 2022 Aug 29. PMID: 36049490; PMCID: PMC9423770

(13) Balmaceda A, Arora S, Sondheimer I, Hollon MM. Resuscitation from a pH of 6.5: A Case Report and Review of Pathophysiology and Management of Extreme Acidosis from Hypovolemic Shock after Trauma. Journal of Trauma and Injury 2019; 32(4): 238-242. Epub 2019 Dec 30. doi: 10.20408/jti.2019.029

(14) Ross SW, Thomas BW, Christmas AB, Cunningham KW, Sing RF. Returning from the acidotic abyss: Mortality in trauma patients with a pH < 7.0. Am J Surg. 2017 Dec;214(6):1067-1072. doi: 10.1016/j.amjsurg.2017.08.033. Epub 2017 Sep 18. PMID: 29079021

(15) Raffee LA, Oteir AO, Alawneh KZ, Alustath AMI. Relationship Between Initial Arterial Blood Gases and Coagulation Profiles – Analyzing the Prognosis and Outcomes in Patients with Multiple Injuries/Trauma. Open Access Emerg Med. 2020 Apr 15;12:87-92. doi: 10.2147/OAEM.S244941. PMID: 32346316; PMCID: PMC7167264.

(16) MacLeod J, Lynn M, McKenney MG, Jeroukhimov I, Cohn SM. Predictors of mortality in trauma patients. Am Surg. 2004 Sep;70(9):805-10. PMID: 15481299

(17) Rudkin SE, Kahn CA, Oman JA, Dolich MO, Lotfipour S, Lush S, Gain M, Firme C, Anderson CL, Langdorf MI. Prospective correlation of arterial vs venous blood gas measurements in trauma patients. Am J Emerg Med. 2012 Oct;30(8):1371-7. doi: 10.1016/j.ajem.2011.09.027. Epub 2011 Dec 12. PMID: 22169587; PMCID: PMC3424304

(18) Boon Y, Kuan WS, Chan YH, Ibrahim I, Chua MT. Agreement between arterial and venous blood gases in trauma resuscitation in emergency department (AGREE). Eur J Trauma Emerg Surg. 2021 Apr;47(2):365-372. doi: 10.1007/s00068-019-01190-6. Epub 2019 Jul 18. PMID: 31321471.

(19) Rudkin SE, Anderson CL, Grogan TR, Elashoff DA, Treger RM. Assessing Acid-Base Status in Circulatory Failure: Relationship Between Arterial and Peripheral Venous Blood Gas Measurements in Hypovolemic Shock. J Intensive Care Med. 2020 May;35(5):511-518. doi: 10.1177/0885066618762335. Epub 2018 Mar 7. PMID: 29514541

(20) Boon Y, Kuan WS, Chan YH, Ibrahim I, Chua MT. Agreement between arterial and venous blood gases in trauma resuscitation in emergency department (AGREE). Eur J Trauma Emerg Surg. 2021 Apr;47(2):365-372. doi: 10.1007/s00068-019-01190-6. Epub 2019 Jul 18. PMID: 31321471.

(21) Patel J, Wiechmann W. Kahn C, Anderson C, Saleem R, Hassan H. 319: Routine Arterial Blood Gases In Trauma Patients: Is It Necessary? Annals of Emergency Medicine: Research Forum abstract| Volume 56, Issue 3, S104, September 2010. doi: 10.1016/j.annemergmed.2010.06.369.

(22) Katirai A, Landau MJ, Berger JM. The utility of abnormal initial arterial blood gas values in determining clinical futility of trauma cases with severe hemorrhage. Am J Emerg Med. 2018 Jul;36(7):1253-1256. doi: 10.1016/j.ajem.2018.03.063. Epub 2018 Mar 23. PMID: 29606404

(23) Barbosa RR, Rowell SE, Diggs BS, Schreiber MA; Trauma Outcomes Group; Holcomb JB, Wade CE, Brasel KJ, Vercruysse G, MacLeod J, Dutton RP, Hess JR, Duchesne JC, McSwain NE, Muskat P, Johannigamn J, Cryer HM, Tillou A, Cohen MJ, Pittet JF, Knudson P, De Moya MA, Schreiber MA, Tieu B, Brundage S, Napolitano LM, Brunsvold M, Sihler KC, Beilman G, Peitzman AB, Zenait MS, Sperry J, Alarcon L, Croce MA, Minei JP, Kozar R, Gonzalez EA, Stewart RM, Cohn SM, Mickalek JE, Bulger EM, Cotton BA, Nunez TC, Ivatury R, Meredith JW, Miller P, Pomper GJ, Marin B. Profoundly abnormal initial physiologic and biochemical data cannot be used to determine futility in massively transfused trauma patients. J Trauma. 2011 Aug;71(2 Suppl 3):S364-9. doi: 10.1097/TA.0b013e318227f170. PMID: 21814105

(24) Vohra T, Paxton J. Abnormal arterial blood gas and serum lactate levels do not alter disposition in adult blunt trauma patients after early computed tomography. West J Emerg Med. 2013 May;14(3):212-7. doi: 10.5811/westjem.2012.3.6905. PMID: 23687538; PMCID: PMC3656700.

(25) Bazzazi A, Hasanloei MA, Mahoori A, Gholamnejad M, Tarverdipour H. Correlation between arterial blood gas analysis and outcome in patients with severe head trauma. Ulus Travma Acil Cerrahi Derg. 2014 Jul;20(4):236-40. doi: 10.5505/tjtes.2014.57089. PMID: 25135016

(26) Carlino MV, Guarino M, Izzo A, Carbone D, Arnone MI, Mancusi C, Sforza A. Arterial blood gas analysis utility in predicting lung injury in blunt chest trauma. Respir Physiol Neurobiol. 2020 Mar;274:103363. doi: 10.1016/j.resp.2019.103363. Epub 2019 Dec 19. PMID: 31866500

(27) Castro D, Patil SM, Keenaghan M. Arterial Blood Gas. [Updated 2022 Sep 12]. In: StatPearls [Internet]. Treasure Island (FL): StatPearls Publishing; 2023 Jan-. Available from: https://www.ncbi.nlm.nih.gov/books/NBK536919/

(28) Kelly AM, Klim S, Rees SE. Agreement between mathematically arterialised venous versus arterial blood gas values in patients undergoing non-invasive ventilation: a cohort study. Emerg Med J. 2014 Oct;31(e1):e46-9. doi: 10.1136/emermed-2013-202879. Epub 2013 Oct 17. PMID: 24136117.

(29) Bijapur MB, Kudligi NA, Asma S. Central Venous Blood Gas Analysis: An Alternative to Arterial Blood Gas Analysis for pH, PCO₂, Bicarbonate, Sodium, Potassium and Chloride in the Intensive Care Unit Patients. Indian J Crit Care Med. 2019 Jun;23(6):258-262. doi: 10.5005/jp-journals-10071-23176. PMID: 31435143; PMCID: PMC6698350.

(30) Burri E, Potocki M, Drexler B, Schuetz P, Mebazaa A, Ahlfeld U, Balmelli C, Heinisch C, Noveanu M, Breidthardt T, Schaub N, Reichlin T, Mueller C. Value of arterial blood gas analysis in patients with acute dyspnea: an observational study. Crit Care. 2011;15(3):R145. doi: 10.1186/cc10268. Epub 2011 Jun 9. PMID: 21663600; PMCID: PMC3219017

(31) Kelly AM. Can VBG analysis replace ABG analysis in emergency care? Emerg Med J. 2016 Feb;33(2):152-4. doi: 10.1136/emermed-2014-204326. Epub 2014 Dec 31. PMID: 25552544.

(32) Kelly AM, Kerr D, Middleton P. Validation of venous pCO2 to screen for arterial hypercarbia in patients with chronic obstructive airways disease. J Emerg Med. 2005 May;28(4):377-9. doi: 10.1016/j.jemermed.2004.10.017. PMID: 15837016.

(33) McCanny P, Bennett K, Staunton P, McMahon G. Venous vs arterial blood gases in the assessment of patients presenting with an exacerbation of chronic obstructive pulmonary disease. Am J Emerg Med. 2012 Jul;30(6):896-900. doi: 10.1016/j.ajem.2011.06.011. Epub 2011 Sep 9. PMID: 21908141.

(34) Brandenburg MA, Dire DJ. Comparison of arterial and venous blood gas values in the initial emergency department evaluation of patients with diabetic ketoacidosis. Ann Emerg Med. 1998 Apr;31(4):459-65. doi: 10.1016/s0196-0644(98)70254-9. PMID: 9546014.

(35) Ma OJ, Rush MD, Godfrey MM, Gaddis G. Arterial blood gas results rarely influence emergency physician management of patients with suspected diabetic ketoacidosis. Acad Emerg Med. 2003 Aug;10(8):836-41. doi: 10.1111/j.1553-2712.2003.tb00625.x. PMID: 12896883.