Throughout this study, we assessed general and clinical knowledge among radiologists, OB/GYN, and EM clinicians regarding best practices in the emergency imaging of pregnant patients. In addition, we hoped to evaluate and improve physicians’ knowledge of the current guidelines surrounding imaging during pregnancy to ensure that patient care isn’t compromised due to fear of radiation.

While on certain questions, performance was strong with over 90% correct answers (#4, #5, #11–13), our study revealed lapses in other domains with less than 50% correct answers on four questions within the questionnaire (#7, 9, 15, and 18). Question #7 of our survey discusses how the radiation exposure for both mother and fetus differs between the following tests: CT pulmonary angiogram (CTPA) and ventilation/perfusion (V/Q) scintigraphy, assuming no dose-modulation techniques were utilized. Current data indicates that the mean maternal effective dose ranges from 0.23 to 9.7 milliSievert (mSv) with CTPA and 0.9 to 5.85 mSv with V/Q lung scanning. The fetal absorbed dose ranges from 0.002 to 0.51 milliGray (mGy) with CTPA and 0.2 to 0.7 mGy with V/Q scintigraphy [17]. In our survey, only 33% of respondents correctly answered this question. Other studies have reported knowledge gaps on this topic as well. In a survey by Groves et al., the authors reported that only 60% of radiologists and a little over half of other clinicians knew that V/Q scintigraphy yields a higher radiation dose to the fetus than CTPA [18]. However, in our survey, some respondents may not have realized the question was framed without dose-modulation techniques. ACR guidelines state that while the radiation dose to the fetus is lower with V/Q scanning, using dose-modulation techniques with CT may make the absorbed dose between the two modalities nearly equivalent [19], and this answer was the second most popular choice, with 29% of responses. Low-dose perfusion-only scanning has also been shown to be similarly efficacious to CTPA while decreasing radiation exposure for the maternal breast, whole body, and fetus compared to CTPA and V/Q [19,20,21,22]. Its similarity to V/Q scanning and its absence amongst the answer choices despite its viability as an alternative may have created confusion, contributing to a low rate of correct responses. These statistics ultimately highlight the fact that educational methodology regarding dosimetry necessitates further review. Furthermore, continual optimization and effective dissemination of protocols are necessary as such advancements occur.

Another question that less than 50% of respondents answered correctly was #18, which asked what imaging techniques could be used to localize a subarachnoid hemorrhage in a pregnant patient. However, only 41.9% of respondents chose magnetic resonance angiography (MRA), one of the correct answers. It is important to note that this discrepancy may have occurred due to the phrasing of the answer choice. MRA may have been interpreted as contrast-enhanced MRA, a test not often utilized in pregnant patients. Perhaps if the answer choice had been written as a non-contrast enhanced MRA (Time-of-flight /phase contrast MRA), clinicians would’ve more readily chosen this answer choice.

Regarding imaging following blunt trauma, for question #8 of the survey, 79% (98/124) of respondents correctly answered that an abdominal/pelvic CT scan was most appropriate for a 34-week pregnant patient after a negative Focused Assessment with Sonography in Trauma (FAST) exam following a motor vehicle accident. Hansen et al. reported similar findings in their study with 88% of radiologists choosing CT as the preferred modality for imaging of a 30-week pregnant patient following an inconclusive initial ultrasound for a patient with abdominal trauma in a motor vehicle accident [23].

In our analysis by training level, significant differences were detected regarding the primary knowledge sources utilized by clinicians. Medical residents predominantly consulted attending physicians/colleagues (7/12, 58%), whereas it was far less common for attendings to do so (15/109, 14%). These results were anticipated based on the fact that trainees are supposed to ask their superiors by design of the medical system. Our study additionally highlights the need for increased clinician and radiologist awareness of established societal practice guidelines to help support clinical decision-making. Just 48% (60/124) of those surveyed drew upon the ACR guidelines as a primary reference source, with utilization lower amongst OB/GYN and EM physicians compared to radiologists. However, we did not directly assess the relationship between knowledge sources and knowledge, and further study in this domain is warranted to develop and target interventions to improve physician awareness of available reference resources.

As far as confidence, most physicians expressed being either “fairly” (58/124, 47%) or “very” (51/124, 41%) confident in making imaging decisions about pregnant patients. Prior reports have also shown that physicians, including trainees, are often imprecise in assigning confidence ratings that align with diagnostic accuracy [24]. In a study by Meyer et al., internal medicine physicians correctly diagnosed 55.3% of uncomplicated cases and just 5.8% of more challenging cases, yet the associated confidence levels were relatively similar (7.2 vs. 6.4 out of 10, respectively) [24]. Further, these authors noted that higher confidence was related to decreased requests for additional diagnostic tests (p = 0.01) [24]. Overconfidence as such may impede clinicians from drawing upon the necessary resources to promote effective clinical decision-making.

One strength of our analysis was the anonymity, which reduces the risk of social desirability bias of answers. Our survey was also distributed internationally, which strengthens the generalizability. Yet, most participants were from the United States, and thus, our results are most applicable to this group. The strength of conclusions drawn from the international participants remains limited by the small sample size. Out of the participants who reported their country of residence, just 14.7% (15/102) were from countries other than the United States. These participants were also spread diffusely amongst such countries, with just 6.9% (7/102) from Canada, 3.9% (4/102) from Hungary, and 1.0% (1/102) from India, Italy, Saudi Arabia, and Slovenia each. In the present study, we did not collect information on the practice setting nor resources available to radiologists. Clinicians in various countries may differ significantly in the resources available to them and the protocols and care team structures, which may have limited the validity of the survey for certain respondents’ practice scenarios. For example, one report from Gujarat state in India found that X-ray and CT services availability was rarely fully adequate due to both the absence and shortage of hardware as well as staffing shortages [25]. These issues were further exacerbated by prolonged equipment breakdowns due to a lack of technicians and engineers [25].

While prior studies on imaging knowledge and practice focused on attending radiologists alone [23], our sample also included OB/GYN and EM physicians and trainees. However, while 108 respondents were radiologists, only 10 EM and 6 OB/GYN physicians participated. Such a sample may have skewed certain results in analyses with stratification by specialty, such as regarding knowledge sources consulted in decision-making, where the ACR guidelines were the most frequently consulted (60/124, 48%). For example, the ACR guidelines were most popular amongst radiologists, the largest group in our sample. EM physicians consulted it as a primary resource less often than radiologists, and OB/GYN clinicians reported not using it as a primary source at all. It is important to note that OB/GYN and EM physicians more likely consult the ACOG or the American College of Emergency Physicians guidelines, which likely explains this disparity. Further, the small sample of OB/GYN and EM physicians limits the strength of our conclusions, such as those noting differences between specialties in decision-making frequency and in general and GI subdomain test results. Similarly, our sample consisted of primarily attending physicians (n = 109), with just three respondents in fellowship and 12 in residency, which impacted our statistical analyses among subgroups. The survey was also lengthy, with a total of 18 questions, and even though 223 clinicians responded, we only had 124 complete responses. Lastly, the physicians who opted to participate in the survey may differ characteristically from those who did not, leading to potential non-response bias.