OC19.02: A novel eye tracking study: how common is expected value bias in fetal growth scan assessment?

Figure 1


Objectives: Operators performing fetal growth scans are usually aware of the gestational age. This may lead to expected‐value bias when measuring biometry, if ultrasound machines display the measurement value on the screen. This study evaluated the occurrence and impact of the expected measurement bias in routine growth scans. Methods: We prospectively collected video recordings of entire routine ultrasound scans coupled with operator eye‐tracking. We defined expected value bias when the operator was looking at the “measurement box” during the process of caliper adjustment, before saving a measurement. We evaluated the occurrence and quantified the direction of bias. Results: We analysed 217 third trimester growth scans performed by 15 operators. Expected value bias occurred in 94% of acquisitions. In 51% vs. 21% of biased acquisitions, the operator adjusted the measurement toward vs. away from the expected value, respectively. On average, measurements were corrected by 2.5 ± 5.7, 2.8 ± 11.3, and 3.2 ± 10.5 days of gestation towards the expected value, for the HC, AC, and FL, respectively (p < 0.001). Comparing the deviation between the measured and expected value, a statistically significant reduction in the variance of HC and FL, but not AC measurements were noted once the operator was biased (P < 0.001 for HC, P = 0.08 for AC, P < 0.01 for FL). Conclusions: During routine obstetric ultrasound scans, expected value bias frequently and significantly changes measurements of standard biometric planes.

Ultrasound in Obstetrics & Gynecology

Table 1
Table 1

*L. Drukker and R. Droste contributed equally to this work.


  author = {Drukker, L. and Droste, R. and Chatelain, P. and Noble, J.A. and Papageorghiou, A.T.},
  title = {OC19.02: A novel eye tracking study: how common is expected value bias in fetal growth scan assessment?},
  journal = {Ultrasound in Obstetrics \& Gynecology},
  volume = {54},
  number = {S1},
  pages = {47-48},
  doi = {10.1002/uog.20550},
  url = {https://obgyn.onlinelibrary.wiley.com/doi/abs/10.1002/uog.20550},
  eprint = {https://obgyn.onlinelibrary.wiley.com/doi/pdf/10.1002/uog.20550},
  year = {2019}