Is radiography or ultrasonography superior at detecting intestinal obstructions in dogs with acute abdominal signs?

[...]

Appraisal, application and reflection

It is essential to quickly and accurately diagnose a surgical condition such as small intestinal obstructions in dogs presenting with acute abdominal signs, as prompt treatment is necessary in providing the best care and prognosis. It is therefore important to understand which modality would be most accurate in detecting the need for surgical intervention, enabling prompt action in investigating cases further. The most common imaging diagnostic tools readily available in first opinion practice are radiography and ultrasonography. This appraisal summarises the findings from six relevant studies to compare the two modalities for when making those decisions for surgical intervention.

Elser et al. (2020) study population included 40 dogs with clinical suspicion of mechanical obstruction and inconclusive initial radiology, presenting with acute abdominal signs. This study has been included for completeness however, the extent to which it answers the question posed in this summary is limited, this is due to multiple issues including the selection criteria of the patients; patients were included in the study only if there was an inconclusive radiographic finding. This places a bias against the usefulness of radiography to diagnose small intestinal obstruction due to all patients with a conclusive radiograph findings being excluded. Ultrasound was used as the contextualised care and radiograph results were referenced against those findings and accuracy was looked at between interpreters rather than overall. There was shown to be a wide range of agreement among the reviewers, based on their experience level. However, the accuracy of correct diagnosis of intestinal obstruction, measured in a mixed animal population (40 dogs and 17 cats) ranged high from 70.2– 89.7% overall through reviewing radiographs.

Winter et al. (2017) compared ultrasound against computerised tomography CT diagnostic imaging, with the contextualised care of exploratory laparotomy for a definitive diagnosis, this allows the accuracy of using ultrasound to be investigated. Ultrasonography showed to have a sensitivity of 100% and specificity of 67% for a correct diagnosis, whereas CT was 100% for both. This shows ultrasound is an extremely useful tool when investigating dogs with acute abdominal signs when CT is not available.

Drost et al. (2016) compared abdominal radiography and CT, for detection of canine mechanical intestinal obstruction, again using exploratory laparotomy as a reference. The sensitivity and specificity of CT on the diagnosis of obstruction were shown to be slightly lower in this investigation at 95.8% and 80.6% respectively, whereas the radiography sensitivity and specificity were 79.2% and 69.4% but the differences were not statistically significant. If directly compared to Winter et al. (2017) ultrasound results against the same referencing standard ultrasound had a higher sensitivity and a higher specificity for diagnosis. However, both studies investigated CT too and found different results in CT specificity and sensitivity. This shows that there were differences, limitations and possible study design problems between the two. Drost et al. (2016) also had the added evaluation of sensitivity and specificity of recommending abdominal surgery after reviewing the radiograph images, this increased the radiographic interpretation with sensitivity rising to 91.7% and specificity to 72.2%. Therefore, even if a definitive diagnosis cannot be given, the important decision of correct intervention is likely to be chosen. It would have been interesting to compare a similar measurement in Winter et al. (2017) methodology using ultrasound in a similar manner if it had been evaluated.

Shanaman et al., (2013) used a cross-sectional design, this study compared both ultrasonography with radiography against exploratory laparotomy and CT in dogs presenting with intense pain and acute abdominal signs. This study did not only focus on small intestinal obstructions but also a wide range of other underlying causes of acute abdominal signs, but it does address the question of which modality was more accurate by comparing them to the contextualised care of CT. The study showed the level of agreement between the modalities in diagnostic accuracy. Both radiography and ultrasonography correctly identified 8 out of 9 surgical cases, with one false negative for each for different reasons, relative to their limitations. Also, there was shown to be good agreement between the two modalities and CT, which was used as the contextualised care, along with abdominal surgery or necropsy. Finally, the study suggests an impression of superiority from abdominal radiographs compared to ultrasound in cases with pneumoperitoneum, but further studies are needed. This is important as intestinal obstructions that caused intestinal ruptures, which are true surgical emergencies, might be better and more quickly diagnosed with radiographs.

Tyrrell & Beck (2006) used a similar approach by also using a cross-section design to their diagnostic validity study, and this study differs from Shanaman et al. (2013) by focusing purely on gastrointestinal foreign bodies not a wider range of underlying conditions. This study looked at the use of radiography versus ultrasonography to diagnose foreign bodies in small animals, and so included cats and dogs in their population. While the PICO of this Knowledge Summary focuses on dogs it was not defined in their radiographic results what the exact number of positive foreign body findings were in the dog population versus cats. Tyrrell & Beck’s (2006) study give clear results on percentages of the accurate diagnoses of the modalities but it does not have any statistical evaluation of comparing them to each other to see if they are significantly different. Out of the 16 cases, of both cats and dogs, radiography identified 56% of the foreign bodies and ultrasound (US) detected 100% of the cases. All of the cases were confirmed through the removal of the material in surgery, endoscopically, or in one case letting it pass naturally in the faeces. While this study supports the findings found in the other studies that US may be more sensitive in detecting foreign bodies in the small intestine in small animals, it does not evaluate its results through statistical analysis, the differences between the modalities are not proven significantly important, the sample is very small, and the risk of bias is high.

Results found in the study by Sharma et al. (2010) were that 58/82 (70.7%) dogs had a definitive result with radiology showing patients being obstructive or unobstructed in the small intestine. And ultrasonography produced definitive results in 80/82 (97.6%) of dogs. And they concluded ‘abdominal ultrasound was more accurate, with fewer equivocal results and provided greater diagnostic confidence compared with radiography, in small intestinal foreign body mechanical obstruction’.

One limitation that most of the studies have are the small sample sizes, with only one mention of a power calculation in which we can apply the results to a wider population (Drost et al., 2016) and one study with a bigger sample population compared to the rest (Sharma et al., 2010). Both Elser et al. (2020) and Tyrrell & Beck (2006) studies had the inclusion of cats into their sample population as well as dogs, this reduces the sample size relevant to this summaries question even further, and so it may be difficult to apply these findings into a much wider population without future studies supporting these results.

There are a variety of levels of attempts of blinding within each study to consider. Winter et al. (2017) had the ultrasound images reviewed in real time and findings were discussed with the attending surgeon, this may have introduced bias based on signalment and clinical signs as well as introducing additional views and potential influence from other clinicians. However, this bias is reduced with the addition of objective measurements used in the assessment, not just subjective views. In the study by Drost et al. (2016), the reviewers were blinded to clinical history and the images were reviewed at least 12 months after the date of the study and viewed in random order. This strengthens the level of blinding and reduces bias. Shanaman et al. (2013) had various people take the radiographs and then interpreters assessed the images, this creates a level of blinding, where the case details were hidden including the signalment and presenting signs of the animal. The primary author was involved through the three assessed modalities, this could have introduced bias as having looked at the previous modality there could be links made between the previous images unconsciously. So instead there may be confirmation of diagnosis rather an independent diagnosis. However, this hypothesis cannot be proven and gets weaker if we consider that the patient history was unknown and the studies were seen retrospectively. Sharma et al. (2010) had a higher level of blinding in that radiographs and ultrasound exams were not done by the interpreters to reduce bias. With Tyrrell & Beck (2006) it was the authors that performed the exams and interpreted them knowing both the patient history and that all animals included in the study had a gastrointestinal foreign body, so there was no blinding possible to the results of the other modality and of the signalment and presenting signs of the case. The radiographs in the study were all viewed by one interpreter and the ultrasound images were analysed by two different people, this may have had an impact on the results due to method and interpretations.

An important limitation of all the studies is the use of highly trained individuals in reviewing the diagnostic images, in comparison to the wider population in general practice. As seen in Elser et al. (2020) study with the use of four reviewers, with a range of experience from first year radiology resident to board-certified radiologist with 16 years experience, there was found to be a wide range of levels of agreement between the reviewers. And this disagreement was with a group of highly trained individuals, since the first year radiology resident had 6 years of veterinary experience including a 1 year diagnostic imaging specialty internship prior to residency. This shows the degree to which experience has an effect on correct interpretation and decision-making, which is an important limitation when extrapolating the results in this summary into the wider population of general practice. However, there were several more problems and limitations in this study that were previously mentioned, that might have affected its results.

In conclusion, it can be seen with the studies evaluated here that both radiology and ultrasound are highly valuable diagnostic tools that can both increase the accuracy of diagnosis and correct decision-making for surgical intervention. It can also be seen that ultrasound has the potential to be superior than radiography at detecting surgical patients with intestinal obstructions in dogs with acute abdominal signs. This potential is highly correlated to the experience of the observer, and so it is recommended that those in general practice should focus on encouraging development and experience in further imaging training, including ultrasound and radiography assessment for patients with acute abdominal signs in order to increase accuracy and confidence in making future surgical intervention decisions. There are major limitations within the studies observed here; focusing mainly on small sample sizes in most of them, which reduces the ability to apply the findings to larger populations, lack of significant differences between radiographic and ultrasound investigations, or even lack of statistical analysis. While there are differences observed the lack of continuous significant differences highlights the fact that there needs to be further research in this area, ideally with larger sample sizes and without the limitations discussed in this summary.

[…]