Alcohol misuse: clinical findings helped diagnose cirrhosis.

Clinical bottom line (level 1b)

  1. Cirrhosis in alcohol abusing men were diagnosed fairly accurately using clinical signs alone.
  2. Clinicians should look for:
    • facial telangiectasia
    • vascular spiders
    • abdominal wall veins
    • white nails
    • fatness
    • peripheral oedema
  3. Patients with no facial telangiectasia and = 2 other signs were at low risk for cirrhosis (< 20%)
  4. Patients were at high risk for cirrhosis (> 80%) if they have:
    • all 6 signs if peripheral oedema
    • 4 or more signs if no peripheral oedema
    • 3 or more signs if facial telangiectasia and no peripheral oedema
  5. Biochemical investigations were unhelpful in diagnosing cirrhosis.
Hamberg et al: Journal of Clinical Epidemiology 1996; 49: 1295-1301
Expires October 2003

The study

Setting: tertiary medical centre, Sweden, 1968-71

303 patients (aged range 31 to 70 years; median 52, 100% male) men drinking >50 g of alcohol per day for > 1 year

Excluded if
  • female
  • unsuitable biopsy
  • previous diagnosis of cirrhosis
  • missing variables

 

Independent blinded reference standard, applied in all patients from a consecutive appropriate spectrum.
Reference standard:
  • liver biopsy
Diagnostic test: clinical findings. History, and physical performed by one senior physician.
  • Logistic regression analysis performed on data.

The evidence

pre-test probability of alcohol abusers: 16.2%, (95% CI: 12% to 20%)
pre-test probability of clinical suspicion of cirrhosis: 27.2%, (95% CI: 20% to 35%)

diagnostic test cirrhosis no cirrhosis LR+
(95% CI)		 LR-
(95% CI)
history of fatigue 34 135 1.3
(1.1 to 1.6) 		0.65
(0.42 to 1.0)
severe dyspepsia 1 15 0.36
(0.05 to 2.6) 		1.0
(0.99 to 1.1)
itching 7 18 2.0
(0.89 to 4.6) 		0.92
(0.82 to 1.0)
previous hepatitis 6 36 0.86
(0.38 to 1.9) 		1.0
(0.91 to 1.2)
alcohol =120 g/day 43 191 1.2
(1.0 to 1.3) 		0.49
(0.23 to 1.1)
abuse =20 years 33 147 1.2
(0.93 to 1.5) 		0.78
(0.51 to 1.2)
fever 4 10 2.1
(0.68 to 6.3) 		0.96
(0.88 to 1.0)
weight loss 18 81 1.2
(0.77 to 1.7) 		0.93
(0.74 to 1.2)
GI bleed 11 41 1.4
(0.77 to 2.5) 		0.92
(0.79 to 1.1)
ascites 5 3 8.6
(2.1 to 35) 		0.91
(0.83 to 1.0)
peripheral oedema 15 48 1.6
(0.99 to 2.7) 		0.86
(0.70 to 1.0)
total 49 254


diagnostic test cirrhosis no cirrhosis LR+
(95% CI)		 LR-
(95% CI)
on examination: jaundice 20 43 2.4
(1.6 to 3.7) 		0.71
(0.56 to 0.90)
encephalopathy 4 3 6.9
(1.6 to 30) 		0.93
(0.85 to 1.0)
fatness 31 102 1.6
(1.2 to 2.0) 		0.61
(0.42 to 0.90)
facial telangiectasia 40 20 10
(6.7 to 16) 		0.20
(0.11 to 0.36)
vascular spiders 28 23 6.3
(4.0 to 10) 		0.37
(0.34 to 0.65)
palmar erythema 31 61 2.6
(1.9 to 3.6) 		0.48
(0.33 to 0.70)
white nails 21 5 22
(8.6 to 55) 		0.78
(0.46 to 0.74)
gynaecomastia 9 8 5.8
(2.4 to 14) 		0.84
(0.74 to 0.96)
ascites 8 3 5.2
(2.0 to 13) 		0.85
(0.76 to 0.98)
enlarged liver 26 30 4.5
(2.9 to 6.9) 		0.53
(0.39 to 0.72)
abdominal wall veins 12 5 12
(4.6 to 34) 		0.77
(0.66 to 0.90)
sparse axillary or pubic hair 12 8 7.8
(3.4 to 18) 		0.78
(0.66 to 0.92)
testicular atrophy 9 8 5.8
(2.4 to 14) 		0.84
(0.74 to 0.96)
peripheral oedema 12 23 2.7
(1.4 to 5.1) 		0.83
(0.70 to 0.98)
total 49 254


diagnostic test cirrhosis no cirrhosis LR
(95% CI)		 post-test probability
logistic regression: facial telangiectasia (FAC) 20.2
(7.0 to 58.9) 		%
logistic regression: vascular spiders (VAS) 6.1
(2.1 to 17.9) 		%
logistic regression: white nails (WHI) 5.8
(1.5 to 22.4) 		%
logistic regression: abdominal wall veins (ABD) 4.4
(0.9 to 20.7) 		%
logistic regression: fatness (FAT) 2.9
(1.1 to 7.8) 		%
total


diagnostic test cirrhosis no cirrhosis LR
(95% CI)		 post-test probability
biochemical doubling: age >40 3.4
(0.9 to 12.9) 		%
biochemical doubling: bilirubin 1.0
(1.3 to 3.4) 		%
biochemical doubling: AST 1.7
(1.1 to 2.5) 		%
biochemical doubling: ESR 1.4
(1.1 to 1.8) 		%
biochemical doubling: clotting factors II, VII, X 0.17
(0.06 to 0.47) 		%
total


diagnostic test cirrhosis no cirrhosis LR
(95% CI)		 post-test probability
logistic regression: facial telangiectasia 19.3
(6.5 to 57.3) 		%
logistic regression: vascular spiders 6.8
(2.3 to 20.1) 		%
logistic regression: white nails 5.0
(1.2 to 20.4) 		%
logistic regression: abdominal wall veins 4.7
(0.9 to 25.0) 		%
logistic regression: fatness 3.7
(1.3 to 10.6) 		%
logistic regression: peripheral oedema 0.22
(0.04 to 1.04) 		%
logistic regression: doubling of alk phos 2.6
(1.1 to 6.2) 		%
total

  • For logistic regression: log odds = -4.18 + 3.01 FAC + 1.80 VAS + 1.75 WHI + 1.48 ABD + 1.07 FAT - 1.28 PER
  • Risk of cirrhosis:
    • Low (<20%)- no facial telangiectasia and 2 or less other signs
    • Moderate- all others combinations
    • High (>80%)- all 6 signs if peripheral oedema; 4 or more signs if no peripheral oedema; 3 or more signs if facial telangiectasia and no peripheral oedema

Comments

  1. There was only one evaluator in this study (a senior resident), and thus there may be considerable variation in interobserver agreement for these findings.

Citation

  1. Hamberg KJ, et al: Accuracy of clinical diagnosis of cirrhosis among alcohol-abusing men. Journal of Clinical Epidemiology 1996; 49: 1295-1301
Search Terms: reference in Bandolier
Contributor: Chris Ball and Bob Phillips, October 2000
Reviewer: Daniel Sontheimer

Clinical Question.

    Patient alcohol abusing men
    Intervention or Exposure clinical signs
    Outcome diagnosis of cirrhosis