We measure shadows, and we search amongst ghostly
errors of measurement for landmarks that are scarcely more substantial.
Edwin Hubble, Astronomer.
Donít lie if you donít have to.
Leo Szilard, Nuclear Physicist.
I can remember the arguments/discussions going
on in our ultrasound room from 1977 on, about how to perform the
bi-parietal diameter (BPD) measurement.
[I was "Student Radiographer" at the time. It was
Trevor Beckwith, Gerald Hayward, Tom Dominikovich and Peter Motteram
having these discussions.] The purists in those days
used to be adamant that a good BPD could only be performed on the
A-mode oscilloscope. The
terminology was "leading edge to leading edge", as we
didnít want to introduce error due to poor axial resolution.
Then there were discussions about which linear echoes on
the B-mode image actually constituted the falx and which were the
septum cavum pellucidum (the WHAT?), or was that the 3rd
ventricle? Someone countered that the landmarks donít actually matter,
as the BPD should be the widest section of the skull wherever that
level may be found. Then
there were those measurements done with a map-reader on Polaroid
prints. Then we had
a character generator (primitive computer) on which we could perform
calibrated measurements. Calibrated
is the word... not! Aspect
ratio, video distortion, light-pen accuracy, etc and so on: these
were part of the everyday potential errors.
A lot of the controversy about this measurement
was derived from its significance in pregnancy dating, and the potential
for assessing growth disturbances later on, plus being a trigger
for the detailed search of fetal malformations, particularly intracranial
stuff. Now we have other methods for assessing much of this.
We look directly at much more of the fetus, so much so that
the 18 week scan has been compared in its completeness to the detail
of an autopsy. We have
various Doppler and AFI and serial AC measurements for growth disturbances.
Is the magic wearing off the BPD?
I hope so. Well, whatever criteria you use, measurement of
the BPD was claimed as being the most accurate measurement of gestational
age known to sonographer-kind with an error of one week to 10 days.
In fact, more and more institutions are instituting (hence
their name) a first trimester scanning policy utilizing the Crown-Rump
length, not only because of the increased accuracy of 5 days to
1 week (and they say thereís no such thing as progress) but also
because of the current measurement du
jour , the nuchal translucency.
This hopefully will make BPD discussions such as this one
as obsolete as mediaeval debate about the number of angels that
could dance on the medulla oblongata of a pin.
Lets us look at some of the causes of error in BPD
measurement, and try to determine whether the lack of a corrected BPD
measurement is a significant cause of fetal death, decreased caesarean
rates (God forbid!), missed golf games, matricide, dogs and cats living
together, other abominations, etc. What
are the causes of error in ultrasound?
This is a highly personal list and should be used in evidence
against me at any time.
Numero uno is measurement botch-ups.
Errors of scan plane selection, faults of measurement obliquity,
sins of caliper placement, minuscule images where the BPD is one
tenth of the screen size (minify the image and you magnify the potential
for error), excess gain ballooning the white bony wchoes, etc...
these are a few of my favourite reasons for wrong-headedness.
(Top pun, that!)
But is this really as bad as I make out?
Is it not inevitable that variations of up to 2mm (1 weeks
or more) are not uncommon even in fastidious scanning in the third
trimester? Have inter-
and intra-observer error studies been performed in every scanning
lab? Are variations
on one patient averaged out or is the second measurement affected
by the first? How accurate are our calipers anyway? Nicolaides only uses 1mm measurement increments in his nuchal
translucency charts to overcome our unconfirmed trust in our machines'
What about the effects of the type of chart you
are using? The ASUM
chart is pretty much accepted in Australia, but some early criticisms
suggested a preponderance of European descent, middle class, Carlton
residents. What about genuine biological variability?
What about variation due to fetal sex?
Fetal position? Does the chart you are using reflect your patient population,
based on racial or other characteristics?
Is your chart based on outer to inner or and outer to outer
measurements? - yes, they exist.
Who actually did the measurements in the study which you
use for your charts - where they experienced sonographers, or first-year
research fellows? Was
it a multi-center study where quality control could be an issue?
Even experienced sonographers make mistakes, but inexperienced
sonographers make a lot more.
Were the charts longitudinal or cross sectional studies? What was the number of data points used?
Were at-risk or abnormal fetuses excluded?
How was the gestational age assessed?
Was the GA averaged up to the nearest whole week (yes, it
happens)? Was the raw
data used to derive the charts? Was the data collected prospectively or retrospectively?
Was the regression curve correctly derived?
Are you using the 5th and 95th centiles,
or two standard deviations?
Head circumference (HC) has been recently shown
to more accurate and reliable than the BPD.
This HC was better if based on direct measurement rather
than calculated from the BPD and OFD (the derived HC).
What does this say?
It might be inferred that the OFD is probably not a reliable
measurement. Most noticeably,
when the same image as the BPD is used to measure the OFD, refractive
shadowing and speed of sound artifacts result in obvious subjectivity
in the placement of the calipers at the occipital end.
Using a second image for OFD has been suggested, a rotated
image where the occiput is well shown, but this raises other points.
Were the techniques used to generate the corrected BPD and
HC charts based on this method of measurement?
If not, can they reasonably be expected to apply?
Also it is hard enough getting people to look properly at
the fetal intracranial structures, let alone do two images just
to get one measurement (the corrected BPD.)
The corrected BPD is actually very close in accuracy to the
derived HC because mathematically they are the same, the difference
only being a multiplying constant.
Therefore using both corrected BPD and derived HC is a redundant
exercise. Also, the
derived HC should be done with an outer to outer BPD, so donít forget
that potential error.
Measurements corrected for head shape certainly
reduce error, but mainly in the later parts of the second trimester and in
the 3rd trimester. For
accurate dating up to the 18week scan, studies have shown that the BPD is
as accurate for practical purposes as the corrected BPD or HC, but with a slightly
wider standard deviation (+/- 8 days as opposed to +/- 7 days.)
Abnormal head shape will certainly affect the BPD measurement, but
is it significant enough to warrant the use of the corrected (however
well) BPD at 18 weeks? Hadlock
does not think so. He
advocates this technique only when the cephalic index (but how reliable is
that?) is more than 1 standard deviation beyond the normal range, but is
aware that it is no better than HC. And
what about head shape where the distortion is superior, like a cone-head
type baby. (Have you seen my baby snaps?) No-one ever corrects for that! [except Terry duBose, except
I am not aware of any studies performed to assess if the
difference in these measurements has any clinical significance.
Most studies have shown that obstetrical ultrasound in general is
an expensive waste of time, and possibly harmful, given the false
reassurances of normality in the woefully performed scans documented in
the RADIUS trial. The
pedantic conclusions of our BPD discussion here are in fact so much
Corrected BPD measurement is in my opinion not worth doing
because the majority of sonographers couldnít do a decent BPD to
begin with if their jobs depended on it (luckily they donít, but
they should). Individual errors of technique are probably far more
significant than those of slight head shape variation (within the
normal range of CI), plus there are the multitude of potential errors
in the charts you are using.
Corrected BPD is the same measurement as derived HC anyway!
The directly measured HC may indeed be the best measurement
after the 2nd trimester, but at 18 weeks it probably
does not make enough difference to alter pregnancy management.
Phillip L Ramm
F. Hadlock, Ultrasound
determination of menstrual age, in Callen P., Ultrasonography in Obstetrics and Gynecology 3rd Ed.
Royston P, Wright EM; How
to construct Ďnormal rangesí for fetal variables. Ultrasound Obstet Gynecol 1998;11:30-38
Altman DG, Chitty L; New
charts for ultrasound dating in pregnancy. Ultrasound Obstet Gynecol 1997;10:192-197
DG, Chitty L Charts of fetal size:
J Obstet Gynaecol 1994;101:29-34