Clinical predictors of bleeding esophageal varices in the ED

Siang-Hiong Goh MBBS*, Wee-Ping Tan MBBS, Shu-Woan Lee MBBS

Accident and Emergency Department, Changi General Hospital, Singapore 529889, Republic of Singapore

Received 23 September 2004; accepted 23 October 2004

Abstract

Objectives: Some authors have found that thrombocytopenia (b118000/mm³), splenomegaly, and ascites are useful predictors of large esophageal varices in Cirrhotic patients. We decide to see whether these factors could also be used to predict bleeding esophageal varices in patients known to have Chronic liver disease in the ED.

Methods: A case record review was done of all patients admitted to the ED of Changi General Hospital with Upper gastrointestinal bleeding from esophageal varices from October 1999 to April 2004. The criteria of thrombocytopenia, splenomegaly, and ascites were applied retrospectively to these patients to see how accurately they performed in predicting bleeding esophageal varices.

Results: Only 55% of patients had thrombocytopenia, whereas 45% had splenomegaly, and 27.5% had ascites. Combining thrombocytopenia with the presence of either ascites or splenomegaly did not improve the yield (only 40%), and only 6 patients had all 3 criteria. Twelve patients with bleeding varices did not have any of the criteria.

Conclusions: Thrombocytopenia, splenomegaly, or ascites is an unreliable predictor of bleeding esophageal varices. Urgent or emergent endoscopy is still advocated to accurately diagnose bleeding esophageal varices.

D 2005

Introduction

It is not uncommon to have patients with chronic liver disease presenting to the emergency department (ED) with acute upper gastrointestinal (GI) bleeding. Often, the approach taken by the emergency physician is to assume that the bleeding originated from esophageal varices and to manage the patient as such until proven otherwise. This is because bleeding esophageal varices have a higher morbid- ity and mortality rate than any other source of upper GI

T Corresponding author. Tel.: +65 8501687; fax: +65 260 3756.

E-mail address: [email protected] (S.-H. Goh).

bleeding [1]. In view of the high hospital mortality rate of 15% to 40% [1], it had been advocated that emergency physicians quickly place such patients on some kind of vasoactive treatment (somatostatin, octreotide, or terlipres- sin [2-5]) while in the ED, in tandem with rapid resuscitation with fluids and blood products. This is to be followed by urgent endoscopy, in which a number of therapeutic options (sclerotherapy [1], variceal multiband ligation [6-8], use of tissue-glue adhesives [6,8]) can be used to stop further esophageal variceal bleeding.

However, 20% to 50% of patients with symptoms and signs of chronic liver disease who develop upper GI bleeding-despite findings such as spider angiomas, asterixis, and ascites-do so from sources other than

0735-6757/$ - see front matter D 2005 doi:10.1016/j.ajem.2004.10.005

esophageal varices [1]. For instance, in the study of Josen et al [9], only 27% of the patients in the liver disease group with varices (cirrhotics) had frank variceal hemor- rhage, whereas 57% bled from hemorrhagic gastritis. Conversely, in approximately 50% of patients with known varices who present with upper GI bleeding, the bleeding is from a source other than the varices [10]. Our own experience in Changi General Hospital showed that for the years 2001 to 2003, only 27% of chronic liver disease patients who presented to the ED with upper GI bleeding did so from esophageal varices. Therefore, starting vasoactive treatment for all patients with chronic liver disease presenting with hematemesis and/or melena (on the assumption that bleeding is from esophageal varices) may not be cost-effective.

Recently, several authors have found that thrombocyto- penia, splenomegaly, and ascites are useful predictors for large esophageal varices in cirrhotic patients [11-18]. We therefore decided to see whether these factors could also be used to predict bleeding esophageal varices in patients known to have chronic liver disease and presenting with hematemesis and/or melena to the ED.

Materials and methods

An electronic search was made of all patients admitted to the ED of Changi General Hospital with an admission diagnosis of GI bleeding (ICD-9 code 578). This search was made from our hospital’s medical record database and began from July 1999 to May 2004. These patients

Results

Based on the search criteria outlined above, a total of 44 patients had been admitted to the ED of Changi General Hospital with upper GI bleeding from esophageal varices between July 1999 and May 2004. The presence of bleeding varices was confirmed in all cases by endoscopy. Four patients were excluded from the study, as 1 had a splenectomy before for hypersplenism, 2 had transjugular intrahepatic portosystemic shunts done before, and another had coexisting Hepatocellular carcinoma. Of the remaining 40 patients studied, 19 had presented to our hospital for the first time. Thirteen (32.5%) were female. The ages of the 40 patients ranged from 29 to 86 years, with a mean age of

55.8 years. Fifteen patients (37.5%) had alcoholism as the etiology of the cirrhosis, 11 (27.5%) patients had hepatitis B, 4 (10%) had Hepatitis C, 1 had Wilson’s disease, 3 patients had dual pathology (alcoholism and hepatitis B or C), with 6 cases being cryptogenic.

Of these 40 patients, the range of platelet counts at the time of presentation to the ED ranged from 38000 to 300000/mm³, with a mean of 129525/mm³, and a median of 117500/mm³. Eighteen patients had Platelet levels equal to or higher than 118000/mm³, both at the time of bleeding in the ED, or in the preceding 3 months at the specialist clinic (although 9 patients had no previous clinic visits).

Table 1 Distribution of platelet levels below 113000/mm³ in 23 patients

were then further selected for the presence of bleeding esophageal varices (ICD-9CM code of 4650). Patients thus chosen for study must have the diagnosis of bleeding esophageal varices confirmed on endoscopy.

	episode	in the preceding 3 months
1	38000	No preceding visit
2	42000	44000
3	46000	No preceding visit
4	61000	No preceding visit
5	67000	No preceding visit
6	67000	65000
7	74000	No preceding visit
8	82000	90000
9	90000	74000
10	90000	62000
11	91000	No preceding visit
12	91000	82000
13	93000	94000
14	93000	No preceding visit
15	95000	89000
16	108000	No preceding visit
17	109000	72000
18	110000	No preceding visit
19	113000	75000
20	116000	80000
21	117000	130000
22	121000	56000

Patients thus identified had their case records reviewed for their initial platelet levels taken at the time of pre- sentation to the ED with upper GI bleeding. If the patient had previous attendances at the specialist clinic or ED within the 3 months preceding the bleeding episode, the baseline platelet levels (if any) was also traced. We identified all those in the 2 groups with a platelet level below 118000/mm³.

We also checked for the presence of ascites and splenomegaly for each patient. This can be from either clinical examination of the abdomen (as detected by the attending physicians) or with the aid of imaging studies (ultrasonography or computed tomography [CT] scan) performed during the same period of hospitalization. The causes of their cirrhosis were also determined. The use of diuretics in patients was also noted, as this could affect the volume of ascites and its clinical detection.

Patients who had splenectomy, splenorenal shunts, or transjugular intrahepatic portosystemic shunts done before were excluded from the study, as were patients with coexisting hepatocellular carcinoma and renal failure.

S/N ED platelet levels (per mm³) at the time of bleeding

Platelet levels (per mm³) obtained from previous specialist clinic or ED visits

thrombocytopenia present
Number with	18	45.0%
splenomegaly present (11 with imaging)
Number with ascites present	11	27.5%
	(5 with imaging)
Number with thrombocytopenia	10	40.0%

Number with all 3 criteria	6	15.0%
Number without any	12	30.0%

Twenty-two patients had platelet counts below 118000/ mm³, either at the time of bleeding in the ED or in the preceding 3 months at the specialist clinic (although 10 of these patients had no previous clinic visits; see Table 1). As such, the criterion was positive in only slightly more than half of the patients (55%; see Table 2).

Table 2 Numbers of patients with the criteria of

thrombocytopenia, splenomegaly, and ascites

Patients with bleeding

esophageal varices Number with

N = 40

55.0%

and 1 other

criteria present (either ascites or splenomegaly)

thrombocytopenia,

splenomegaly or ascites

Splenomegaly was detected on physical examination in 7 patients, and imaging studies (either ultrasonography or CT scan) showed the presence of splenomegaly in an additional 11 patients. In total then, 45% (18) of patients had splenomegaly (Table 2).

Ascites was detected on physical examination in

6 patients, and imaging studies (either ultrasonography or CT scan) showed the presence of ascites in an additional

5 patients. In total then, 27.5% (11) of patients had ascites. We would like to point out that 5 of 15 specialist clinic follow-up patients who had no ascites were on spironolactone treatment at the time of the bleeding episode (Table 2).

Table 3 Summary of various papers citing clinical predictors for esophageal varices in cirrhotic patients

^a Detected on ultrasonography.

^b Detected by either clinical examination or imaging (ultrasound or CT scan).

^c Detected on ultrasonography or CT scan.

Finally, only 6 patients with bleeding had all 3 criteria of thrombocytopenia, splenomegaly, and ascites, whereas 12 patients had none of the 3 criteria (Table 2).

Discussion

Several authors [11-18] have put forward papers suggest- ing that thrombocytopenia, splenomegaly, and ascites are useful predictors for large esophageal varices in cirrhotic patients. This is intuitive as the triad is the result of portal hypertension in cirrhotic patients. These criteria were developed for the purpose of improving yield and cost- effectiveness when selecting patients for screening endos- copy and prophylactic therapy to decrease the incidence of bleeding. A brief summary of the salient points of these studies is listed in Table 3, along with the references. Interestingly, the authors found that the Child-Pugh score was not a reliable predictive criterion for esophageal varices. We used 40 patients who had confirmed bleeding esophageal varices by endoscopy (taking endoscopy as the bgold standardQ) and applied the criteria of thrombocytope- nia, splenomegaly, and ascites to this group to see how

accurately they could predict bleeding varices.

Levels of platelet counts recommended for the screening of esophageal varices by various authors [11-15] varied from 150000/mm³ (proposed by Freeman et al [15]) to 68000/mm³ (proposed by Madhotra et al [12]). However, Freeman et al [15] limited their study population to only alcoholic cirrhotic patients. For the purpose of our study, therefore, we applied the next highest platelet cutoff level of 118000/mm³ (Table 3), which was proposed by Thomo- poulos et al [11], whose study population included all forms of liver cirrhosis. We included platelet levels done both at the time of bleeding in the ED and any platelet levels done within the preceding 3 months (usually by the relevant discipline in the specialist outpatient clinics). Using this criterion, only 55% of patients with bleeding esophageal varices fulfilled it.

Author	Year	Number of	Factor studied by authors
		patients studied	Thrombocytopenia (cutoff level)	Splenomegaly	Ascites
Thomopoulos et al [11]	2003	184	Yes (118000/mm³)	Yes^a	Yes^a
Madhotra et al [12]	2001	184	Yes (68 000/mm³)	Yes^b	Yes^b
Chalasani et al [13]	1999	346	Yes (88 000/mm³)	Yes^b
Zaman et al [14]	1999	98	Yes (88 000/mm³)	Yes^b
Freeman et al [15]-study limited	1999	65	Yes (150 000/mm³)
to alcoholic cirrhotic patients only Barcia et al [16]	1998	95			Yes^c
Alacorn et al [17]	1998	79	Yes (not specified)	Yes^b	Yes^b
Lavergne et al [18]	1997	52			Yes^a

The criterion of splenomegaly was advocated in papers by Thomopoulos et al [11], Madhotra et al [12], Chalasani et al [13], Zaman et al [14], and Alacorn et al [17]. The last

4 authors used both clinical examination and imaging studies to detect the presence of splenomegaly, whereas Thomopoulos et al [11] used imaging studies. This is understandable as clinical examination alone to detect a mild-to-moderately enlarged spleen is often unreliable and subject to many variables such as the experience of the clinician, position of the patient (lying on the right as compared to lying supine), the presence of ascites, and the abdominal girth of a patient. In our series, only 7 patients had a palpably enlarged spleen on clinical examination at the time of the variceal bleeding, whereas imaging done within the same period of hospitalization revealed an additional 11 patients with splenomegaly. All in all, only 18 patients (40.0%) fulfilled the criteria of splenomegaly.

In the same vein, the criterion of ascites was advocated in papers by Thomopoulos et al [11], Madhotra et al [12], Barcia et al [16], Alacorn et al [17], and Lavergne et al [18]. Madhotra et al [12] and Alacorn et al [17] used both clinical examination and imaging studies to confirm the presence of splenomegaly, whereas Thomopoulos et al [11], Barcia et al [16], and Lavergne et al [18] used both imaging studies. The detection of free fluid in the abdomen by clinical examination depends on relatively crude techniques (percussion for shifting dullness, and fluid thrill) [19]. At least 500 to 1000 mL is required to produce abnormal physical signs, whereas ultrasound can detect as little as 50 mL [19]. In our series, only 6 patients had ascites on clinical examination at the time of the variceal bleeding, whereas imaging done within the same period of hospital- ization revealed an additional 5 patients with ascites. All in all, only 11 patients (27.5%) fulfilled the criteria of ascites. The use of spironolactone would have reduced the number of patients with ascites. (Some might argue against using imaging to detect splenomegaly and ascites in patients with bleeding varices. However, an increasing number of emergency physicians are using emergency ultrasonogra- phy for trauma to detect fluid in the traumatized abdomen. In the future, the experience thus gained can facilitate the bedside detection of splenomegaly and ascites in patients in the ED.)

Combining thrombocytopenia with either the presence of ascites and splenomegaly did not improve the yield (only 40%), and only 6 patients had all 3 criteria. We would also have missed 12 patients with bleeding varices, none of whom had any of the criteria. As the percentages were low, statistical analysis for tests of significance was not attempted.

One limitation was the small number of patients (40) in this study, which might lead to a type II error, but this represented all the patients seen in our hospital in the last

4 years with confirmed bleeding esophageal varices by endoscopy. Perhaps another study with a larger number of patients could be done, although we doubt whether any differences will arise.

Conclusion

Some authors have suggested that thrombocytopenia, splenomegaly, and ascites are useful predictors for large esophageal varices in cirrhotic patients and can be used for the purpose of improving yield and cost-effectiveness when electively selecting patients for screening endoscopy and prophylactic therapy to decrease the incidence of first-time variceal bleeding.

However, these 3 factors fail to translate into reliable predictors of bleeding esophageal varices in chronic liver disease patients presenting with hematemesis and/or melena to the ED. Urgent or emergent endoscopy is still advocated for such patients to accurately diagnose bleeding esophageal varices and institute appropriate treatment.

References

McQuaid KR. Alimentary tract. In: Tierney Jr LM, McPhee SJ, Papadakis MA, editors. Current medical diagnosis and treatment, 43rd ed. New York (NY)7 Lange Medical Books/McGraw-Hill; 2004.

p. 515 - 622.

Gow PJ, Chapman RW. Modern management of oesophageal varices. Postgrad Med J 2001;77:75 - 81.
de Franchis R. Somatostatin, somatostatin analogues and other vasoactive drugs in the treatment of bleeding oesophageal varices. Dig Liver Dis 2004;36(Suppl 1):S93 - S100.
Goulis J, Burroughs AK. Role of vasoactive drugs in the treat- ment of bleeding oesophageal varices. Digestion 1999;60(Suppl 3): 25 - 34.
D’Amico G. The role of vasoactive drugs in the treatment of oesophageal varices. Expert Opin Pharmacother 2004;5:349 - 60.
Helmy A, Hayes PC. Current endoscopic therapeutic options in the management of variceal bleeding. Aliment Pharmacol Ther 2001; 15:575 - 94.
Nevens F, Rutgeerts P. Variceal band ligation in the management of bleeding oesophageal varices: an overview. Dig Liver Dis 2001; 33:284 - 7.
Lata J, Hulek P, Vanasek T. Management of acute variceal bleeding. Dig Dis 2003;21:6 - 15.
Josen AS, Giuliani E, Voorhees Jr AB, et al. Immediate endoscopic diagnosis of upper gastrointestinal bleeding. Its accuracy and value in relation to associated pathology. Arch Surg 1976;111:980 - 6.
Erdman CW, Mensching JJ. Gastrointestinal bleeding. In: Stone CK, Humphries RL, editors. Current emergency diagnosis and treatment, 5th ed. New York (NY)7 Lange Medical Books/McGraw-Hill; 2004.

p. 283 - 95.

Thomopoulos KC, Labropoulou-Karatza C, Mimidis KP, et al. Non- invasive predictors of the presence of large oesophageal varices in patients with cirrhosis. Dig Liver Dis 2003;35:473 - 8.
Madhotra R, Mulcahy HEI, Willner I, et al. Prediction of esophageal varices in patients with cirrhosis. J Clin Gastroenterol 2002;34:81 - 5.
Chalasani N, Imperiale TF, Ismail A, et al. Predictors of large esophageal varices in patients with cirrhosis. Am J Gastroenterol 1999;94:3285 - 91.
Zaman A, Hapke R, Flora K, et al. Factors predicting the presence of esophageal or gastric varices in patients with advanced liver disease. Am J Gastroenterol 1999;94:3292 - 6.
Freeman JG, Darlow S, Cole AT. Platelet count as a predictor for the presence of oesophageal varices in alcoholic cirrhotic patients. Gastroenterology 1999;116:A1211 [abstract].
Barcia HX, Robalino GA, Molina EG, et al. Clinical predictors of large varices in cirrhotic patients. Gastrointest Endosc 1998; 47:A78.
Alacorn F, Burke CA, Larive B. Esophageal varices in patients with cirrhosis: is screening endoscopy necessary for everyone? Am J Gastroenterol 1998;93:1673 [abstract].
Lavergne J, Molina E, Reddy KR, et al. Ascites predicts the presence of high grade varices by screening endoscopy. Gastrointest Endosc 1997;45:A649.
Gray D, Toghill P. The gastrointestinal tract. An introduction to the signs and symptoms of clinical medicine. London (UK)7 Arnold; 2001.

p. 213 - 25.

AJEM