Vomiting in infancy is a common problem and often is difficult to solve on clinical grounds alone. Without resolution, these infants eventually are referred to radiology, and in this regard, they may be imaged with an upper GI series, ultrasound of the stomach, or nuclear scintigraphy. In this article, the authors discuss the kind of information each study provides, and which test to use for a given clinical scenario.
is a Fellow in the Department of Radiology and
is Professor of Radiology and Pediatrics and Director of
Pediatric Radiology, both at the University of Texas Medical
Branch in Galveston, TX.
omiting in infancy is a common problem and often is difficult to
solve on clinical grounds alone. Usually the first therapeutic
measure consists of changing the infant's formula, but many times
multiple changes are required, and the infant continues to vomit.
Without resolution these infants eventually are referred to a
radiologist for evaluation, and in this regard, the following
imaging choices are available: 1) an upper GI series, 2) ultrasound
of the stomach, and 3) nuclear scintigraphy. However, before any of
these studies are ordered, one should be aware of what kind of
information each provides and when each should be used.
What causes vomiting?
--There are many causes of vomiting, and while the majority are
mechanical in origin and referable to the gastrointestinal tract,
vomiting also can occur with certain metabolic diseases, neurologic
diseases, and infections such as sepsis, pneumonia, and meningitis.
Inflammatory diseases of the gastrointestinal tract also cause
vomiting, and the most common problem is gastroenteritis, usually
of a viral origin. In other cases vomiting can be due to
obstructions secondary to problems such as intussusception, delayed
midgut volvulus, and underlying congenital bands or stenoses, but
overall, the most common cause of chronic vomiting in infancy is
gastric in origin and centers around pylorospasm, pyloric stenosis,
and gastritis. Infants with these conditions can be examined with a
variety of imaging modalities, and exact choices and sequences may
vary from institution to institution. In our institution, however,
ultrasound of the stomach, and specifically the antrum, has proven
to be the most useful initial, and usually only, study to be
performed. Indeed, it is the mainstay of the investigation of these
patients for it is the most direct way to identify the underlying
If vomiting is present, does one need to demonstrate
--Very often a radiologist will receive a request for an imaging
study, usually an upper GI series, with the following inscription:
"patient vomiting, spitting up, please rule out reflux". This,
however, is a contradiction, for if the patient is vomiting,
regurgitating, or spitting up, then the patient is refluxing.
Therefore, there is no need to perform an upper GI series or any
other imaging study to demonstrate that this is occurring. It is
more important to determine why the patient is refluxing. In other
words, is there a gastric outlet or duodenal obstruction, or is the
problem simple chalasia with a normal antrum? The latter is
diagnosed after exclusion of a gastric or duodenal obstruction, and
in our institution ultrasound has proven most useful for
accomplishing this task.
All of this is not to infer that other well known imaging
studies should never be performed. On the contrary, nuclear
scintigraphy (most sensitive), followed by the upper GI series both
can identify occult reflux such as might occur with chronic
aspiration, apneic spells, SIDS, etc.
What diseases may be encoun-tered?
--For the most part, the following causes of vomiting in the infant
patient may be encountered: 1) a normal antrum with chalasia, 2)
pyloro-spasm, 3) pyloric stenosis, 4) gastritis, and 5) delayed
congenital obstructions of the stomach and duodenum. Some of these
conditions lend themselves to the use of one specific imaging
modality over others, but overall, most initially can be evaluated,
and usually accurately diagnosed, with ultrasound. Indeed,
ultrasound exams which are now available with high resolution (7 to
10 MHz) linear transducers allow for the evaluation of the gastric
antrum to the level of depiction of the individual layers of the
gastrointestinal tract. At the same time, the radiologist is able
to study the physiology of gastric emptying and, thus, ultrasound
is an ideal study for the evaluation of chronic vomiting in
Upper GI series
The upper GI series has the inherent problem of delivering
ionizing radiation to the patient. This is not an insurmountable or
deadly problem, though it is one that should be avoided, if
possible. In the past, the upper GI series was the only study
available and although it is quite good at demonstrating gastric
outlet obstruction, delayed gastric emptying, and duodenal
obstructions, the antral findings often were nonspecific. This
commonly occurred with pylorospasm and pyloric stenosis but was
less of a problem with gastritis. The upper GI series is much
better at identifying problems such as the rare gastric diaphragm
or the slightly more common duodenal diaphragm.
Ultrasound of the stomach
In our institution, as mentioned earlier, ultrasound has become
the most commonly performed imaging procedure in the evaluation of
infants with persistent vomiting. It yields accurate data as to the
status of the antrum and also gives an accurate perception of
gastric emptying. However, to be useful it must be performed with
high resolution (7 to 10 MHz) linear transducers. When such
transducers are used, not only is the antrum visualized, but so are
all the layers of the gastric wall. Ultrasound also can demonstrate
gastroesophageal reflux, though this should not be the focus of the
The diagnostic criteria for the ultrasonographic diagnosis of
pyloric stenosis consist of a fixed, spastic, and elongated
antropyloric canal associated with thickening of the circular
Initially, much attention was paid to the length and
cross-sectional diameter of the pyloric canal. Indeed even
volumetric measurements have been suggested, but most currently
rely on measurement of the actual thickness of the pyloric muscle.
There is no question that the pyloric canal is elongated in these
patients, but the finding is so obvious that it is not worth
pursuing in terms of specific measurement. Nonetheless, the
diagnostic length of an elongated pyloric canal initially was said
to be 1.7 cm or greater, but this measurement has dropped to around
1.2 cm. In reality, however, all of this is moot, for the diagnosis
of pyloric stenosis, for the most part, is easily made with simple
inspection. The configuration of the elongated pylorus, along with
the thickened muscle, is so persistent that it leaves little doubt
about the diagnosis (figure 1). Even measuring thickness of the
muscle, if the observer is experienced, is not necessary.
Measurements are of value only in borderline cases.
In terms of muscle thickness, 4 mm or greater initially was
thought to be diagnostic of pyloric stenosis, though it
subsequently was proven that 3 mm or greater is sufficient.
Lesser degrees of thickening are not consistent with pyloric
stenosis, and these patients should not be treated surgically. On
the other hand, 2 to 3 mm of thickness should not be considered
normal, and indeed, such thicknesses frequently can be seen with
pylorospasm, with or without associated gastritis (figure 3). These
patients have abnormal gastric emptying and will require treatment.
For the most part this is accomplished medically with an
antispasmodic agent such as metachlopromide or bentyl.
The classic findings of pyloric stenosis, as noted earlier, are
relatively easy to identify (figure 1). The elongated pyloric canal
with thickened muscle is rather characteristic. However, one or two
pitfalls exist. The most common of these is echogenicity at the
6:00 and 12:00 cross/sectional positions which, when seen on a
longitudinal section in the mid-sagittal plane, can cause the
muscle to appear echogenic and almost invisible (figure 2). This
echogenicity has been found to be due to reflections from the
interfaces of the circular muscle fibers as they pass through the
6:00 and 12:00 positions.
However, it also can result from a reverberation artifact. At any
rate, once appreciated, it is of no real consequence.
The other major pitfall deals with posterior positioning of the
pyloric canal. Characteristically in pyloric stenosis, the pyloric
canal demonstrates some degree of posterior positioning and
curvature. However, when the stomach is full of food or fluid, as
is often the case in these infants, it can hide the posteriorly
positioned pyloric canal, making it possible to miss the thickened
muscle mass entirely. This was a problem in the early stages of
ultrasound, but, with experience, it has now been resolved. Once
these potential pitfalls are appreciated, the ultrasonographic
diagnosis of pyloric stenosis, in experienced hands, approaches
100%. Indeed, from a practical standpoint it is almost foolhardy to
endlessly try to palpate an olive in these patients if it is not
palpable immediately. Ultrasound is so direct and definitive that
it can save much time and effort.
Findings of pylorospasm are similar to those of pyloric
stenosis, but while the pyloric canal initially may be elongated
and fixed, it does not remain fixed permanently (figure 3).
Eventually with pylorospasm the antropyloric canal opens and
peristaltic activity passes through it. In many of these cases the
pyloric muscle may be slightly thickened, but it is never thickened
beyond 2 mm. If it is thickened to 3 mm or over, the patient should
be diagnosed with pyloric stenosis rather than pylorospasm.
If one adheres to these criteria, the diagnosis of pylorospasm
is relatively straightforward and the findings easy to interpret.
Pylorospasm is treated medically with the administration of
antispasmodic drugs such as metachlopromide. This leads to
relaxation of the antrum and normalization of gastric emptying.
Most patients respond to this form of therapy, but in some cases,
even if treated medically, progression to classic pyloric stenosis
can be seen.
Gastritis also is relatively easily identified with ultrasound,
and in infants it is more common than generally appreciated.
Most often, it is secondary to milk allergy. On ultrasound
examination, the mucosa appears thickened and echogenic and
measures well over the normal 2 to 3 mm
; thickening may be circumferential or eccentric (figure 4).
Duodenal obstructions, such as duodenal bands with midgut
volvulus (figure 5A), also can be identified with ultrasound. The
main finding in these cases is a distended descending duodenum.
Similar findings are more vividly demonstrable with upper GI
series, including spiraling of the small bowel along the twisted
mesentery (figures 5B,5C).
In this regard, ultrasound has become valuable in demonstrating
reversal of the normal relationship of the superior mesenteric
artery and vein (figure 6). Although not foolproof, reversal of the
position of these two vessels is highly suspicious and virtually
diagnostic of malrotation.
In addition, if volvulus is present, mesenteric fat and the
mesentery itself, along with the mesenteric vessels, can be seen to
form a concentric ring-like configuration leading to the so-called
"whirlpool" sign (figure 6).
Ultrasonography also can demonstrate gastric diaphragms (figure
7A). However, many times these abnormalities are more vividly
demonstrable with an ordinary upper GI series (figure 7B).
Nuclear scintigraphy has been found to be the least useful study
for the evaluation of chronic vomiting. While it provides
information as to rate of gastric emptying and the presence of
reflux, it lacks specificity regarding the various antral problems
encountered. In terms of gastric emptying, unless the study is
performed properly (e.g., with the patient positioned on his right
side or held upright) the findings can be misleading. If the study
is performed with the patient on his back though in a semi-erect
position, erroneous data may be obtained. Often in such cases the
dilemma of the ultrasound study (or in former years, the upper GI
series) demonstrating normal gastric emptying while the nuclear
scintigraphy study suggests significant and prolonged delay in
Nuclear scintigraphy has no specificity as far as the type of
obstructive problem present in the antrum. Therefore, when it is
performed as a first study and is abnormal, it is still absolutely
necessary to perform another type of diagnostic study (usually
ultrasound in our institution). As this is not in the least cost
effective, nuclear scintigraphy is best used for the detection of
occult reflux, as seen in patients with apneic spells, reactive
airway disease with occult reflux, and SIDS.
Chronic vomiting in infancy is a common problem and is often
difficult to solve clinically. With imaging, specifically the
ultrasound study, a definitive diagnosis can be accomplished in
virtually every case. This being the case, it is most cost
effective to proceed to this study immediately and not perform the
other studies instead.