Observer Variability Exists in the Assessment of Fatty Liver on Ultrasound

The objective of a recent study was to examine the variability between interobserver and intraobserver sonographic assessment of a fatty liver.
The study has concluded that subjective visual assessment of fatty liver on ultrasound is quite variable between observers as well as among the same observer

Participants
In the retrospective analysis, 168 patients who had undergone ultrasound examination of the abdomen and qualified for study according to the following criteria:
technically adequate ultrasound study, homogeneous liver echotexture, no focal hepatic lesion, and at least 1 image included both the right kidney and liver.

Methodology
Images were reviewed by 3 radiologists with experience in abdominal sonography. Each case was graded as follows:

• normal
• mild hepatic steatosis
• moderate hepatic steatosis
• or severe hepatic steatosis.

Mild fatty liver was seen with a mild increase in liver echogenicity. Moderate fatty liver was seen with impaired visualization of the intrahepatic vasculature and diaphragm.
Severe fatty liver was seen with markedly increased hepatic echogenicity, decreased penetration of the posterior segment of the right lobe of the liver, and reduced or no visualization of the hepatic vasculature and diaphragm.
An initial interpretation was performed. Approximately 4 weeks later, cases were randomly rearranged in a new sequence, and observers were requested to assess the presence and severity of fatty liver under the same conditions and without knowledge of their initial interpretation.
The interobserver and intraobserver agreement percentages were calculated by the number of complete agreements by dividing the total number of readings. The degree of agreement was based on weighted kappa statistics and interpreted as follows:
<0.00 was poor agreement, 0.00 to 0.20 was slight agreement, 0.21 to 0.40 was fair agreement, 0.41 to 0.60 was moderate agreement, 0.61 to 0.80 was substantial agreement, and 0.81 to 1.00 was almost perfect agreement.

The mean interobserver percentage of agreement for assessing presence or absence of hepatic steatosis was approximately 76%.

The interobserver agreement for presence as well as severity of fatty liver between first and second observer, first and third observer, and second and third observer was at a moderate level at both first and second interpretations.

Reviewer’s Comments
These Results are helpful in demonstrating the considerable variability in the subjective visual assessment of presence and severity of fatty liver on ultrasound, not only between different interpreters but also by the same interpreter.

One limitation in this study was that the subjective visual assessment was not performed as real-time but rather with static images.

Author: : John C. Sabatino, MD, MSD

Reference: :
Strauss S, Gavish E, et al. Interobserver and Intraobserver Variability in the Sonographic Assessment of Fatty Liver.
AJR; 2007; 189 (December): W320-W323