Original article

Role of liver function tests in predicting common bile duct stones in acute calculous cholecystitis
W. K. Peng, Z. Sheikh, S. Paterson-Brown and S. J. Nixon
University Department of Clinical and Surgical Sciences (Surgery), Royal Inrmary of Edinburgh, Little France Crescent, Edinburgh EH16 4SA, UK Correspondence to: Mr S. J. Nixon (e-mail: Stephen.Nixon@ed.ac.uk)

Background: The role of liver function tests (LFTs) in evaluating common bile duct (CBD) stones in patients with cholelithiasis has been studied widely. However, it is not clear whether these predictive models are useful in inammatory gallstone disease. Methods: A review was undertaken of 385 consecutive patients admitted as an emergency for acute calculous gallbladder disease. The diagnosis of calculous cholecystitis was conrmed by ultrasonography or histological conrmation of acute or chronic inammation of the gallbladder. Patients with obvious jaundice, dened as a bilirubin level above 80 mol/l, and gallstone pancreatitis were excluded. Results: Some 216 patients met the inclusion criteria, of whom 28 (130 per cent) were found to have CBD stones. LFT results were not signicantly different in patients with chronic, acute or complicated acute cholecystitis. Using several cut-off levels, -glutamyl transpeptidase (GGT) had the highest specicity, positive predictive value and negative predictive value, comparable to a scoring system that combined all LFTs. Bilirubin was the least specic and predictive. A cut-off point for GGT at 90 units/l produced a sensitivity of 86 per cent (24 of 28), specicity of 745 per cent (140 of 188), and positive and negative predictive values of 33 per cent (24 of 72) and 972 per cent (140 of 144) respectively. This represented a one in three chance of CBD stones when the GGT level was above 90 units/l and a one in 30 chance when the level was less than 90 units/l. Conclusion: Selection criteria based on GGT can be used in acute calculous cholecystitis to identify high-risk patients who would benet most from further imaging to exclude choledocholithiasis.

Paper accepted 3 March 2005 Published online 3 August 2005 in Wiley InterScience (www.bjs.co.uk). DOI: 10.1002/bjs.4955

Introduction

Common bile duct (CBD) stones are present in 1125 per cent of patients with gallstone-associated gallbladder disease1 6 , but risk factors identied before surgery do not accurately predict their presence. As a result some surgeons adopt a policy of routine intraoperative cholangiography (IOC) during laparoscopic cholecystectomy, whereas others tend to be more selective. Likewise, indications for endoscopic retrograde cholangiopancreatography (ERCP) vary and the possible benets of magnetic resonance cholangiopancreatography (MRCP) are being evaluated7 9 . The situation in acute cholecystitis is further complicated by the necessity for rapid investigation before early cholecystectomy and by the increased technical difculties faced by surgeons when performing an operation associated with a higher conversion rate than in the absence of acute inammation. To add to the difculty,
Copyright 2005 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

liver function test (LFT) results in acute cholecystitis are often deranged owing to the inammatory process, and it is not clear whether LFTs remain useful in identifying patients with inammatory gallstone disease at high risk of having CBD stones. Most previous studies have assessed LFTs in patients with symptomatic cholelithiasis who are undergoing cholecystectomy1 3,9 17 . Few have evaluated their role in acute inammatory gallstone disease; the results of most of these studies were inconclusive and routine IOC was advocated4 6,18 21 .

Patients and methods

Details of all patients admitted to the general surgical unit of the Royal Inrmary of Edinburgh are collected prospectively using the Lothian Surgical Audit System.
British Journal of Surgery 2005; 92: 12411247

1242

W. K. Peng, Z. Sheikh, S. Paterson-Brown and S. J. Nixon

Between January 2000 and December 2001, 385 consecutive patients with acute calculous gallbladder disease were referred by the accident and emergency unit or directly by their general practitioner to the emergency surgical team; their data were reviewed retrospectively. The data system records all subsequent admissions, thereby allowing follow-up for a minimum of 1 year in this study and a maximum of 2 years after the acute admission for identication of patients with a late presentation of CBD stones. Diagnosis of inammatory gallstone-related gallbladder disease was based on ultrasonographic criteria (gallbladder wall thickness greater than 3 mm with or without oedema, pericholecystic uid, etc.) or histological conrmation of acute or chronic inammation. The cohort in this study therefore included patients with acute or acuteon-chronic calculous cholecystitis and those with chronic cholecystitis presenting as acute biliary pain. Patients with obvious clinical jaundice, dened as a bilirubin level above 80 mol/l, and gallstone pancreatitis were excluded. IOC was performed according to each consultants working practice. The rst LFT results during the acute admission were used for analysis. LFT parameters studied were: bilirubin (normal range 217 mol/l), alanine aminotransferase (ALT; normal range 1040 units/l), alkaline phosphatase (ALP; normal range 40125 units/l) and -glutamyl transpeptidase (GGT; normal range 535 units/l).

Table 1 Derangement of liver function in the presence of common bile duct stones Common bile duct stones First LFT results in acute admission Bilirubin (mol/l) ALT (units/l) ALP (units/l) GGT (units/l) Present (n = 28) 26 (369) 191 (9590) 189 (24700) 347 (151767) Absent (n = 188) 16 (377) 55 (41021) 95 (20543) 87 (91067)

P* 0012 < 0001 < 0001 < 0001

Values are mean (range). LFT, liver function test; ALT, alanine aminotransferase; ALP, alkaline phosphatase; GGT, -glutamyl transpeptidase. *Wilcoxon test.

Table 2 Comparison of liver function in patients without common bile duct stones who had emergency surgery for chronic, acute and complicated acute cholecystitis Cholecystitis without CBD stones First LFT results in acute admission Bilirubin (mol/l) ALT (units/l) ALP (units/l) GGT (units/l) Complicated acute cholecystitis (n = 20) 15 (631) 33 (488) 82 (22177) 62 (12194)

Chronic cholecystitis (n = 60) 16 (470) 69 (81021) 87 (20543) 85 (91067)

Acute cholecystitis (n = 57) 13 (336) 36 (5420) 87 (42207) 56 (10449)

P* 0380 0134 0460 0172

Statistical analysis
The results of all LFT parameters were tabulated in descending order and then plotted graphically. A cut-off value with optimal specicity and sensitivity was obtained for each parameter, and a scoring system was created by combining these optimal cut-off values for each LFT parameter. Any result above the cut-off value scored 1 point, the minimum total score being 0 and the maximum 4. In addition, an imaginary line was plotted to represent a perfect parameter. The data were analysed statistically for two and three unpaired quantitative data using non-parametric Wilcoxon sum-of-ranks and KruskalWallis tests respectively. A two-tailed Fisher exact test was used for qualitative data. P < 0050 was considered statistically signicant.
Results

Values are mean (range). LFT, liver function test; CBD, common bile duct; ALT, alanine aminotransferase; ALP, alkaline phosphatase; GGT, -glutamyl transpeptidase. *KruskalWallis test.

Of the 385 patients who were admitted as an emergency with acute gallstone disease, 216 fullled the inclusion criteria; 140 (648 per cent) were female patients and the
Copyright 2005 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

mean age was 54 (range 1499) years. Some 124 patients (574 per cent) had evidence of acute inammation and 92 (426 per cent) had chronic inammation. Of the 216 patients, 193 underwent cholecystectomy (181 attempted laparoscopic cholecystectomies, of which 25 (138 per cent) were converted, and 12 open cholecystectomies); 154 were operated on during the same admission and 39 during a subsequent elective admission. The remaining 23 patients did not have surgery, but remained well within the follow-up period. CBD stones were found in 28 patients (130 per cent); 22 were detected during the acute admission, two during subsequent elective admissions and four presented after cholecystectomy. Biliary imaging (ERCP, MRCP or IOC) was performed in 110 patients. There was a difference in the incidence of CBD stones between patients aged over 55 years (19 of 100; 190 per cent) and those aged less than 56 years (nine of 116; 78 per cent) (P = 0016, Fishers exact test). No difference was seen in the incidence of CBD stones between
www.bjs.co.uk British Journal of Surgery 2005; 92: 12411247

Liver function tests and common bile duct stones

1243

male and female patients (ten of 76 versus 18 of 140 respectively; P = 1000, Fishers exact test). In the presence of CBD stones, the mean values of ALT, ALP and GGT were increased (P < 0001, Wilcoxon test) (Table 1), and for bilirubin (P < 0050). After excluding patients with common duct stones, comparison of LFT results in patients operated on during acute admission showed no signicant difference between those with histological evidence of chronic, acute or complicated acute cholecystitis (Table 2). The severity of inammation in the gallbladder did not, therefore, inuence the LFT results. The results for each liver function parameter for all patients were arranged in descending order (Table 3). For each series of results, all the patients with CBD stones (from the rst to the last, or 28th, patient) were identied and their respective positions in the series were plotted (Fig. 1). For each series, a cut-off point taken at the rst patient would identify only one patient with CBD stones (sensitivity 1/28 100 per cent), whereas cut-off at the last patient would identify all patients with stones (100 per cent sensitivity). The specicities and positive predictive values (PPVs) at these cut-off values were calculated and plotted in Figs 2 and 3 respectively. An additional line representing an imaginary test with perfect specicity was drawn. This is a straight line

Table 3 Results of each liver function test arranged in descending order of concentration Order of LFT result 1 2 3 4 5 7 8 9 10 11 12 13 14 17 18 19 20 21 22 23 Bilirubin (mol/l) 77 73 70 69* 66* 58* 56* 56 50* 44* 43 43 42* 42 41 41* 40 40 39 38 ALT (units/l) 1021 590* 589 588* 526* 468* 456* 456 441* 420 371 325* 291 288* 286 273* 264 247 238 234 ALP (units/l) 700 543 508* 411 405 380* 342* 326* 321* 277 250* 247 237* 226 207 192 185 180 177 175* GGT (units/l) 1767* 1101 1067 849 732* 654 645 616* 611* 575* 551 483 479 468* 462 449 424* 399 362 341*

Only the rst 23 of the 216 results for each parameter are shown. Patients at a particular position for one parameter are likely to have different positions for other parameters. *Patients with common bile duct stones. LFT, liver function test; ALT, alanine aminotransferase; ALP, alkaline phosphatase; GGT, -glutamyl transpeptidase.

216

201 Position in LFT re sults order

151

Perfect Bilirubin ALT ALP GGT Scoring system

101

51

1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th 11th 12th 13th 14th 15th 16th 17th 18th 19th 20th 21st 22nd 23rd 24th 25th 26th 27th 28th Patients with choledocholithiasis

Fig. 1

Patients with common bile duct stones and their position in the liver function test (LFT) results order. The horizontal axis indicates patients with conrmed stones according to their ranking in Table 3. ALT, alanine aminotransferase; ALP, alkaline phosphatase; GGT, -glutamyl transpeptidase

Copyright 2005 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

British Journal of Surgery 2005; 92: 12411247

1244

W. K. Peng, Z. Sheikh, S. Paterson-Brown and S. J. Nixon

100 90 80 70 Specificity (%) 60 50 40 30 20 10

0

Perfect Bilirubin ALT ALP GGT Scoring system

1/28 2/28 3/28 4/28 5/28 6/28 7/28 8/28 9/28 10/28 11/28 12/28 13/28 14/28 15/28 16/28 17/28 18/28 19/28 20/28 21/28 22/28 23/28 24/28 25/28 26/28 27/28 28/28

Test sensitivity ( 100%)

Fig. 2 Test specicity at different cut-off values. ALT, alanine aminotransferase; ALP, alkaline phosphatase; GGT, -glutamyl transpeptidase

100

90 Perfect Bilirubin 80 ALT ALP 70 Positive predictive value (%) GGT Scoring system

60

50

40

30

20

10

1/28 2/28 3/28 4/28 5/28 6/28 7/28 8/28 9/28 10/28 11/28 12/28 13/28 14/28 15/28 16/28 17/28 18/28 19/28 20/28 21/28 22/28 23/28 24/28 25/28 26/28 27/28 28/28

Test sensititvity ( 100%)

Fig. 3

Positive predictive value at different cut-off values. ALT, alanine aminotransferase; ALP, alkaline phosphatase; GGT, -glutamyl transpeptidase

Copyright 2005 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

British Journal of Surgery 2005; 92: 12411247

Liver function tests and common bile duct stones

1245

representing all patients with CBD stones occupying the top 28 positions in the series (Fig. 1). This line represents an imaginary indicator with 100 per cent specicity (Fig. 2). Each LFT initially formed an almost straight line close to the perfect indicator, before deviating away from this perfect line (Fig. 1). The further the line deviated from the perfect indicator, the lower the test specicity obtained; thus, to obtain higher test sensitivity, the specicity would drop (Fig. 2). GGT had the closest t to the perfect test line, but the curve for bilirubin showed an abrupt breakdown as early as the 12th patient (Fig. 1). At a higher test sensitivity, GGT maintained better specicity and PPV than other LFTs (Figs 2 and 3). Bilirubin was the least specic and predictive compared with other LFTs (Figs 2 and 3). For each series, a cut-off point (Fig. 1) was identied before the graph line broke down. To obtain a test specicity of at least 70 per cent and a PPV of at least 30 per cent, the cut-off points chosen were 31 mol/l (12th patient) for bilirubin, 59 units/l (19th patient) for ALT, 112 units/l (19th patient) for ALP and 90 units/l (24th patient) for GGT. The specicity and PPV produced by the scoring system were close to that for GGT alone (Figs 13). When similar graphs were produced for patients for whom biliary imaging had indicated the presence or absence of CBD stones with certainty, the results were the same, with GGT having the best specicity. A cut-off point for GGT at 90 units/l produced a sensitivity of 86 per cent (24 of 28), specicity of 745 per cent (140 of 188), and positive and negative predictive values of 33 per cent (24 of 72) and 972 per cent (140 of 144) respectively. One-third of patients (72 of 216) with cholecystitis had a GGT level above 90 units/l and a one in three chance (24 of 72) of having CBD stones. Patients with a GGT level of less than 90 units/l had approximately a one in 30 chance (four of 144) of CBD stones.
Discussion

For each LFT parameter in this study, a similar pattern emerged: when levels were greatly increased, all parameters were specic in predicting the presence of bile duct stones, closely following the line of a perfect predictive test. However, the line for bilirubin showed an early breakdown from the perfect line, indicating that at a lower cut-off value bilirubin is no longer specic in predicting CBD stones. By contrast, GGT maintained an overall better specicity and PPV than all the other LFTs, comparable to that of a combined scoring system. The decision had been taken to exclude patients with a bilirubin level
Copyright 2005 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

above 80 mol/l as these patients would have signicant clinical jaundice and therefore already be regarded as high risk. The management of CBD stones is currently in a state of change with the realization that small asymptomatic stones may pass spontaneously, the introduction of MRCP and the growing use of laparoscopic stone extraction7 9,22 24 . Preoperative indicators of CBD stones have been studied widely in the past, but with variable results1 6,9 21 , hence routine IOC has been proposed to detect bile duct calculi5,6,15,17,18 . However, routine IOC presents the surgeon with a dilemma if immediate laparoscopic exploration and stone extraction cannot be performed. The bile duct calculi then need to be retrieved by postoperative ERCP or conversion to open surgery. Although ERCP is a reliable method for imaging the biliary tree, it is invasive and carries its own risk of complications. Conversion to open surgery to allow bile duct exploration is associated with increased morbidity in terms of longer hospital stay and recovery period25 . As not all the patients in this study had denitive biliary imaging, leading to some uncertainty about the presence of CDB stones, the results were analysed using data from only those patients who had denitive CBD imaging; the same overall result was found, in support of GGT as the most specic and predictive parameter. Although CBD stones cannot be excluded without biliary imaging, the fact that patients remained well during follow-up suggests the absence of any clinically signicant ductal calculi, and spontaneous migration of small duct stones is an established phenomenon22,23 . In the past, other LFTs have been studied widely in preference to GGT, perhaps because of the reluctance of researchers to use this indicator, which can be inuenced by alcohol consumption. Although GGT was omitted from most previous studies of LFTs, those that did include it showed GGT to be the most sensitive marker of common duct stones11,20 . Alcohol intake in the present patients was not recorded accurately and could not be assessed retrospectively, but, if known, might increase the predictive value of GGT in low consumers. Studies of CBD stones and their effect on liver function are difcult to evaluate as choledocholithiasis is a dynamic disease. ERCP may be carried out based on deranged liver function, only to nd a clear bile duct or some evidence of the recent passage of a stone. It is anticipated that deranged liver function remains unchanged for a variable period of up to 6 weeks after complete biliary drainage26 . This explains why deranged LFT results do not always accurately predict the nding of bile duct stones, as these may have been present but passed by the time of biliary
www.bjs.co.uk British Journal of Surgery 2005; 92: 12411247

1246

W. K. Peng, Z. Sheikh, S. Paterson-Brown and S. J. Nixon

imaging, and bile duct stones have been known to migrate spontaneously22,23 . Recent evidence has recommended MRCP as the procedure of choice in the preoperative identication of CBD stones7 9 , enabling referral to a surgical team with appropriate expertise in laparoscopic duct exploration. In the authors unit, like many in the UK, where expertise in laparoscopic CBD exploration is not routinely available to all surgeons, MRCP is increasingly used before surgery to identify patients with CBD stones, thereby reducing the need for diagnostic ERCP and limiting this potentially high-risk investigation to patients who require a therapeutic intervention. In the ideal situation, MRCP should be performed in all patients, but this is an expensive option. In smaller hospitals, MRCP may not be available at all, and in larger units it may still be difcult to obtain MRCP within 48 h, as required in the ideal management of acute cholecystitis. It is therefore still relevant for many surgeons in the UK with limited access to emergency MRCP to identify patients at high risk of CBD stones, including those with ultrasonographic evidence of stones or biliary dilatation, jaundice, gallstone pancreatitis1,2,9,16,21 or, as shown by this study, a GGT level above 90 units/l. A GGT cut-off level of 90 units/l identies one in three patients with cholecystitis who require biliary imaging, the remainder having a very low risk of having CBD stones. The authors intend to apply this policy to patients presenting with acute inammatory gallstone disease and audit its effect, with the expectation of a reduced hospital stay, better planning of common duct stone management and a decreased need for ERCP. With the increasing use of early laparoscopic cholecystectomy in acute-onset gallstone disease, a rapid method for identifying patients with bile duct stones is required. This study has demonstrated that LFTs can be used to predict bile duct stones, even when the gallbladder is inamed. All LFTs are of some discriminatory value, but a raised GGT level is the most specic and sensitive, and as good as a scoring system based on all four routine LFTs. In addition to known risk factors such as ultrasonographic evidence of CBD stones or biliary dilatation, jaundice and gallstone pancreatitis1,2,9,16,21 , the authors propose that a GGT level greater than 90 units/l also indicates a high risk of CBD stones, and that these patients should undergo more detailed examination of the common duct by MRCP, ERCP or IOC, depending on unit policy. Patients with a GGT concentration below 90 units/l, and no other risk factor, have a one in 30 chance of having a CBD stone and could be treated expectantly.
Copyright 2005 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

References
1 Menezes N, Marson LP, de Beaux AC, Muir IM, Auld CD. Prospective analysis of a scoring system to predict choledocholithiasis. Br J Surg 2000; 87: 11761181. 2 Onken JE, Brazer SR, Eisen GM, Williams DM, Bouras EP, DeLong ER et al. Predicting the presence of choledocholithiasis in patients with symptomatic cholelithiasis. Am J Gastroenterol 1996; 91: 762767. 3 Koo KP, Traverso LW. Do preoperative indicators predict the presence of common bile duct stones during laparoscopic cholecystectomy? Am J Surg 1996; 171: 495499. 4 Jarvinen H. Abnormal liver function tests in acute cholecystitis; the predicting of common duct stones. Ann Clin Res 1978; 10: 323327. 5 Carlsen JE, Lauritzen T, Juul K, Hermann C, Hansen PH. Common duct stones in patients with acute cholecystitis. Acta Chir Scand 1977; 143: 4748. 6 Stryker SJ, Beal JM. Acute cholecystitis and common-duct calculi. Arch Surg 1983; 118: 10631064. 7 Taylor AC, Little AF, Hennessy OF, Banting SW, Smith PJ, Desmond PV. Prospective assessment of magnetic resonance cholangiopancreatography for noninvasive imaging of the biliary tree. Gastrointest Endosc 2002; 55: 1722. 8 Kats J, Kraai M, Dijkstra AJ, Koster K, Ter Borg F, Hazenberg HJ et al. Magnetic resonance cholangiopancreaticography as a diagnostic tool for common bile duct stones: a comparison with ERCP and clinical follow-up. Dig Surg 2003; 20: 3237. 9 Topal B, Van de Moortel M, Fieuws S, Vanbeckevoort D, Van Steenbergen W, Aerts R et al. The value of magnetic resonance cholangiopancreatography in predicting common bile duct stones in patients with gallstone disease. Br J Surg 2003; 90: 4247. 10 Kiechle FL, Weisenfeld MS, Karcher RE, Epstein E. Alkaline phosphatase in the assessment of choledocholithiasis before surgery. Am J Emerg Med 1985; 3: 556560. 11 Pereira-Lima JC, Jakobs R, Busnello JV, Benz C, Blaya C, Riemann JF. The role of serum liver enzymes in the diagnosis of choledocholithiasis. Hepatogastroenterology 2000; 47: 15221525. 12 Del Santo P, Kazarian KK, Rogers JF, Bevins PA, Hall JR. Prediction of operative cholangiography in patients undergoing elective cholecystectomy with routine liver function chemistries. Surgery 1985; 98: 711. 13 Barkun AN, Barkun JS, Fried GM, Ghitulescu G, Steinmetz O, Pham C et al. Useful predictors of bile duct stones in patients undergoing laparoscopic cholecystectomy. McGill Gallstone Treatment Group. Ann Surg 1994; 220: 3239. 14 Huguier M, Bornet P, Charpak Y, Houry S, Chastang C. Selective contraindications based on multivariate analysis for operative cholangiography in biliary lithiasis. Surg Gynecol Obstet 1991; 172: 470474. 15 Cranley B, Logan H. Exploration of the common bile duct the relevance of the clinical picture and the

www.bjs.co.uk

British Journal of Surgery 2005; 92: 12411247

Liver function tests and common bile duct stones

1247

16

17

18 19

20

21

importance of peroperative cholangiography. Br J Surg 1980; 67: 869872. Abboud PA, Malet PF, Berlin JA, Staroscik R, Cabana MD, Clarke JR et al. Predictors of common bile duct stones prior to cholecystectomy: a meta-analysis. Gastrointest Endosc 1996; 44: 450455. Saltzstein EC, Peacock JB, Thomas MD. Preoperative bilirubin, alkaline phosphatase and amylase levels as predictors of common duct stones. Surg Gynecol Obstet 1982; 154: 381384. Pitluk HC, Beal JM. Choledocholithiasis associated with acute cholecystitis. Arch Surg 1979; 114: 887888. Norrby S, Heuman R, Sjodahl R. Acute cholecystitis. Frequency of stones in the common duct and predictive value of liver function tests. Ann Chir Gynaecol 1985; 74: 912. Hammarstrom L, Ranstam J. Factors predictive of bile duct stones in patients with acute calculous cholecystitis. Dig Surg 1998; 15: 323327. Reiss R, Deutsch AA, Nudelman I. Clinical signicance of choledochal diameter and hyperbilirubinemia in acute

cholecystitis. Int Surg 1985; 70: 129132. 22 Frossard JL, Hadengue A, Amouyal G, Choury A, Marty O, Giostra E et al. Choledocholithiasis: a prospective study of spontaneous common bile duct stone migration. Gastrointest Endosc 2000; 51: 175179. 23 Collins C, Maguire D, Ireland A, Fitzgerald E, OSullivan GC. A prospective study of common bile duct calculi in patients undergoing laparoscopic cholecystectomy: natural history of choledocholithiasis revisited. Ann Surg 2004; 239: 2833. 24 Rojas-Ortega S, Arizpe-Bravo D, Marin Lopez ER, Cesin-Sanchez R, Roman GR, Gomez C. Transcystic common bile duct exploration in the management of patients with choledocholithiasis. J Gastrointest Surg 2003; 7: 492496. 25 Buanes T, Mjaland O. Complications in laparoscopic and open cholecystectomy: a prospective comparative trial. Surg Laparosc Endosc 1996; 6: 266272. 26 Watanapa P. Recovery pattern of liver function after complete or partial surgical biliary decompression. Am J Surg 1996; 171: 230234.

Copyright 2005 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

British Journal of Surgery 2005; 92: 12411247