Lung ultrasound training and evaluation for proficiency among physicians in a low-resource setting

Lung ultrasound (LUS) is an effective tool to evaluate patients with dyspnea in the emergency department (ED) [1‐5]. Guidelines recommend including LUS during point-of-care ultrasound training in emergency medicine [6]. As an operator-dependent skill, LUS requires adequate training in order to be used effectively [7]. In a recent systematic review of LUS training, Pietersen et al. recommends a three-step approach of teaching theoretical knowledge, followed by hands-on sessions on simulators or healthy subjects, and finally supervised scanning of patients to determine when learners are ready for independent scanning [8]. However, there is no standardized recommendation on the amount of training required or the number of examinations needed to achieve proficiency in LUS [8]. Guidelines for LUS competency are based on expert opinion and lack prospective data to support them [9].

Only a handful of prior studies have evaluated LUS proficiency in learners [10‐15]. Several of these studies primarily focused on learners in the intensive care unit (ICU) with varied supervision and assessments of proficiency [10, 14, 15]. While Arbelot et al. included ED residents, the study was performed in the ICU setting [15]. Other studies primarily focused on particular disease processes, such as pneumothoraces [13] or patients with cardiogenic pulmonary edema [12]. There remains a lack of evidence regarding the amount of training and the number of LUS examinations needed to attain proficiency for evaluating the undifferentiated dyspneic patient presenting to the ED. This is of particular importance in a resource-limited setting where radiographs are harder to obtain. The objective of this study was to determine the threshold number of LUS examinations physicians need to perform to achieve proficiency for interpreting LUS in ED patients with dyspnea.

Twenty-one physicians were enrolled in the study; however, two physicians withdrew from the study after leaving the hospital for personal reasons, leaving nineteen physicians who completed the study. The majority of physicians in the study were medical officers (84%) who had less than 1 year clinical experience following medical school (Table 1). Most had never used an ultrasound machine or had used the ultrasound less than ten times. The general practice faculty and two emergency medicine fellows (general practice physicians doing an 18-month training in emergency medicine) had used the ultrasound a large amount, primarily performing obstetrical ultrasounds. None of the study physicians had performed LUS prior to the study.

A total of 330 lung examinations were performed with 3288 lung zones included in analysis. Physicians submitted an average of 17 complete lung ultrasound scans (SD 5.5, range 10–30). Several examinations were missing posterior lung views due to patient’s clinical status and inability to sit up or roll over. One hundred and sixteen (35%) lung examinations were normal and 214 (65%) had abnormalities. The most common diagnosis for dyspnea in this ED population was pneumonia, one of the most common causes of morbidity and mortality in resource-limited settings (Table 2).

Eighteen physicians (95%) reached proficiency. Physician S only provided 10 scans and did not reach proficiency in this analysis. We found that physicians attained proficiency for interpreting LUS accurately when compared to an expert after 4.4 (SD 2.2) LUS studies for individual zone interpretation and 4.8 (SD 2.3) LUS studies for overall interpretation; see Figs. 1, 2. Table 3 provides a summary of statistics for the number of scans required to reach an acceptable level of training by physician.

Physicians also demonstrated good image quality with an average image quality of 4.1 (SD 0.3). The image quality improved as more scans were performed.

Expert interobserver agreement for individual images was 0.7 and for overall diagnosis was 0.9.

This was the first study to evaluate the number of LUS scans needed to attain proficiency for acquiring and interpreting images on undifferentiated dyspneic ED patients in a resource-limited setting. Overall, we found that nearly all novice users became proficient in acquiring and interpreting lung ultrasounds after performing on average less than five independent examinations when compared to an expert.

These results are similar to the study by See et al., evaluating proficiency in lung ultrasound use by respiratory therapists in an intensive care setting [10]. In their study, respiratory therapists only needed ten directly observed scans to accurately obtain and interpret images. However, these exams were all directly observed and may provide an added confidence that may not be there when scanning independently. This study did not follow learners after supervised scans to evaluate if proficiency was maintained while scanning independently. In our study, novice physicians scanned independently and were able to attain proficiency within the same number of scans.

Millington et al. developed and evaluated a tool for assessment of competency in thoracic ultrasonography on ten learners, finding learners rapidly improved in image generation and interpretation in up to 25–30 scans [11]. However, the study did not evaluate sequential ultrasounds from learners or define when competency was achieved. Similarly, a training curriculum and prospective evaluation by Arbelot et al. recommended 20–25 supervised scans to acquire basic skills for interpretation of LUS. [15] While our study provided five supervised scans, we found proficiency was obtained after only an additional five scans performed independently. This difference may be due to differences in the complexity of ICU versus ED patients, which highlights the need for further evaluation of proficiency needs for learners in different settings.

Russell et al. found that novice sonographers, including physicians and non-physicians, achieved proficiency in LUS B-line quantification after limited training and independently scanning 11 lung zones (less than two full LUS examinations) [12]. This study assessed 29 learners and used Cusum analysis to construct learning curves [12]. It differs from our study in that they specifically evaluated proficiency for B-line quantification in acute heart failure patients instead of determining LUS findings and diagnosis in ED patients with undifferentiated dyspnea. The data from both of these studies remain consistent with prior research suggesting that LUS is one of the easier ultrasound exam types to perform and interpret [22].

In our study, one learner did not reach proficiency. Soon after enrollment into the study, the learner transitioned from the ED to a clinical research position. This transition may have resulted in less vested interest in learning ultrasound. Additionally, the learner tried to quickly complete the number of ultrasounds, which may have made it difficult to implement any feedback to develop proficiency. While all other learners obtained proficiency within ten scans, this highlights the importance that some learners may require more scans.

There are several limitations within this study. We had a small sample size and conducted the study within one center, making generalizability difficult. Also, there may be a selection bias as physicians volunteered to participate in the study, which may lead to those physicians with more motivation to learn ultrasound being included in the study. Additionally, we used the BLUE protocol for interpretation of LUS findings. The BLUE protocol was developed in the ICU setting on critically ill patients, and does not differentiate subtle findings (i.e., interstitial infections from edema or pneumonia from other causes of consolidation). However, the BLUE protocol provides a helpful algorithm for novice users in learning how to apply lung ultrasound to clinical care. Because of this, we utilized this organized approach in our training and encouraged learners to interpret these findings within the clinical context. We did not collect body mass index (BMI) or other patient demographics for the study as this is difficult to do in a resource-limited setting. BMI may impact image quality and thus interpretation. Although lung ultrasound is considered easy to acquire [22], future studies should assess impact of BMI on proficiency. Finally, interpretation is dependent on clinical context and we did not follow patients to ensure that the clinical diagnosis matched the LUS interpretation. It is important to recognize that with LUS it is relatively easy to learn how to acquire images; however, interpretation of artifacts has a steeper learning curve. In this study, we found that learners achieved proficiency for interpretation after just five independent scans. When applying LUS clinically, it is important to remember that findings and subsequent management of the patient should be done within the clinical context.

Overall, this study further helps to understand LUS proficiency and may be used to help guide future benchmark recommendations. Having a better understanding of the number of scans necessary to reach proficiency is crucial to determine the level of training needed for providers using LUS. Future studies evaluating ongoing proficiency and impact on clinical care would be helpful, particularly in resource-limited settings where other diagnostic tools are lacking.

Following 8 h of didactic and hands-on training, the majority of physicians novice to lung ultrasound achieved proficiency with interpretation of lung ultrasound after less than five ultrasound examinations performed independently.