Pericardiocentesis is a life-saving emergent procedure. Though performed rarely by emergency physicians, it is a skill they must know how to perform expertly when necessary. It is a critical skill taught and learned in all emergency medicine residency programs. Pericardial effusions can develop from medical causes (post-viral, related to cancer, infection, or post-surgical) or from traumatic injury to the heart. When a pericardial effusion impairs the filling of the heart, termed pericardial tamponade, hemodynamic instability ensues. This is an immediately life-threatening condition that must be treated with pericardiocentesis. Ideally performed in a controlled, surgical setting by an interventional cardiologist, this procedure is performed in the emergency department when the patient is rapidly decompensating and other resources are not available.

Traditionally performed using anatomical landmarks via a subxiphoid approach, the increasing and nearly ubiquitous use of point-of-care cardiac ultrasonography by emergency physicians has made ultrasound-guided pericardiocentesis in the emergency department a possibility. This procedure is best performed under ultrasound guidance, as the complication rate is much lower (0.5%-3.7%)^[1,2,3,4] when compared to blind or electrocardiography-assisted pericardiocentesis (15%-20%).^[5,6] Potential complications include right ventricular puncture, pneumothorax, gastric puncture, hepatic puncture, life-threatening hemorrhage, and resulting infection.

When using ultrasound guidance, the approach where the largest fluid collection visualized on ultrasound can be accessed closest to the skin surface is recommended. Ultrasound allows for a physician to utilize entry sites other than the subxiphoid area, such as a parasternal approach or an apical approach. To our knowledge, there is no clear consensus in the literature which approach is ideal. The expert- and consensus-based recommendations for needle placement and entry sites for pericardiocentesis vary. The Advanced Cardiac Life Support (ACLS) manual recommends a parasternal approach as the safest.^[7] This approach is just lateral to the sternum on the anterior chest. The American College of Surgeons Advanced Trauma Life Support (ATLS) manual recommends a subxiphoid approach.^[8] The most authoritative text on emergency medicine procedures recommends using the approach with which the clinician is most familiar, stating that the subxiphoid approach is the most common.^[9] Tintinalli's Emergency Medicine: A Comprehensive Study Guide recommends a subxiphoid approach if the procedure must be performed blind.^[10] A Mayo clinic study characterized all of the pericardiocenteses performed at their institution over a 21-year period; 1,127 pericardiocenteses were performed by cardiologists and of these, 208 were performed via subcostal or subxiphoid approach (18% of all total procedures) and 83 were performed parasternally (7% of all total procedures). The largest proportion from this study was performed from an apical position (n=714, 63%).^[11] The apical position is not described or recommended in any of the emergency medicine literature, ACLS, or ATLS recommendations. In another study which retrospectively reviewed pericardiocentesis performed in the intensive care unit for patients with tamponade physiology, the subxiphoid approach was used 99% of the time (109 of 110 pericardiocentesis performed).^[12] We have been unable to find any studies characterizing which approach is most frequently used in practice for emergency department pericardiocentesis, likely because this is such a rare emergency department procedure. Given the conflicting recommendations, our goal was to investigate the ideal entry site for point-of-care ultrasound-guided pericardiocentesis in the emergency department. The aim of this study is to identify the ideal approach for emergency-physician-performed ultrasound-guided pericardiocentesis as determined by ultrasound image quality, distance from surface to pericardial fluid, and likely obstructions or complications.

This was a retrospective review of point-of-care cardiac ultrasound examinations performed in two urban academic emergency departments with a combined annual census of 120,000. Both emergency departments have an Accreditation Council for Graduate Medical Education (ACGME)-accredited emergency medicine residency and an active point-of-care ultrasound training program. The emergency medicine residents receive point-of-care ultrasound training and milestone assessment according to ACGME guidelines. Hospital-based credentialing in point-of-care ultrasound is available per American College of Emergency Physicians (ACEP) guidelines,^[13] and quality assurance is performed for all examinations by emergency ultrasound fellowship-trained physicians. All point-of-care ultrasound examinations are archived in the web-based workflow solution database, Q-path (Telexy Healthcare, Port Coquitlam, Canada), which stores images and interpretation reports. An Institutional Review Board (IRB) approval was obtained for this study.

Eligible point-of-care cardiac ultrasound examinations were identified through search of the emergency department ultrasound image archival system (Q-path, Telexy Healthcare, Port Coquitlam, Canada) during the study period. Emergency medicine residents and attending physicians with varied point-of-care ultrasound experience performed cardiac ultrasound examinations as part of routine patient evaluation after initial clinical assessment. The examinations were performed with a Z.One Ultra (Zonare Medical Systems, Mountain View, CA, USA) or a SonixTouch (Ultrasonix Medical Corporation, Richmond, British Columbia, Canada) system with a low-frequency broadband phased array transducer. Emergency physicians with varying ultrasound examination skills performed the studies, ranging from first-year residents to attending physicians.

All point-of-care ultrasound studies performed in the emergency departments over the study period were reviewed for the presence of cardiac images. Patients were included in the study if they received a point-of-care cardiac ultrasound examination, were over 18 years old, or were found to have a pericardial effusion (Figure 1). All images were reviewed by an emergency ultrasound fellowship-trained physician for technical quality, distance of effusion from skin surface to pericardial fluid, obstructions to the predicted path of the needle, and predicted complications. The technical quality of the images was rated on a grading scale from 1-5 using the proposed ACEP suggested quality assurance grading scale.^[14] The distance from skin surface to pericardial fluid was measured from saved images using manual measurements and the recorded sonographic scale. The size of the pericardial effusion was recorded and definitions were standardized based on the criteria proposed by Armstrong.^[15] Obstructions to the predicted path of the needle or predicted complications were recorded. These included obstruction by another organ such as the lung, stomach, or liver, poor image quality or prohibitive depth. The investigators used their professional judgement as to whether obstructions, prohibitive depth, or predicted complications existed.

The primary outcomes were predicted complication rates for pericardiocentesis for different cardiac views and the mean skin-to-pericardial fluid distance. The secondary outcomes included image quality ratings for different cardiac views.

Descriptive statistics are presented along with 95% confidence intervals (95% CI). Proportions were compared using Fisher's exact test, means were compared using the Student's t-test, and medians were compared using the Mann-Whitney U-test. All analyses were conducted in STATA 15 (College Station, Texas).

A total of 166 pericardial effusions were identified. Subxiphoid, parasternal, and apical views were each obtained for 84.3% (95% CI 77.9%-89.5%), 73.5% (95% CI 66.1%-80.0%), and 60.2% (95% CI 52.4%-67.7%) of examinations, respectively. The presence of a pericardial fluid collection was visualized in each view as follows: 88.6% (95% CI 82.1%-93.3%) of subxiphoid views, 81.1% (95% CI 73.0%-87.7%) of parasternal views, and 72.0% (95% CI 62.1%-80.5%) of apical views (P=0.005). Image quality ratings were similar for the three views, with a median and interquartile range (IQR) rating of 3 (3-4) for the subxiphoid views, 4 (3-5) for the parasternal views, and 4 (3-4) for the apical views. The mean skin-to-pericardial fluid distance was significantly greater for the subxiphoid views compared to both the parasternal and apical views (Table 1). Each view was evaluated for the potential complications. The subxiphoid view had the highest predicted complication rate at 79.7% (95% CI 71.5%-86.4%) and was significantly greater than the apical (31.9%; 95% CI 21.4%-44.0%, P<0.001) and parasternal (20.2%; 95% CI 12.8%-29.5%, P<0.001) views. The most frequently predicted complication for the three views was the liver puncture in the subxiphoid views (83.7%; 95% CI 74.8%-90.4%), poor image quality precluding safe procedure in the parasternal view (30.0%; 95% CI 11.9%-54.3%), and inability to access the fluid collection in the apical view (26%; 95% CI 10.2%-48.4%). The specific frequencies of predicted complications for the subxiphoid view are summarized in Table 2.

Our study is relevant in a variety of settings where pericardiocentesis is performed, such as in the emergency department, intensive care unit, and other acute care environments. Pericardial effusions are increasingly detected on point-of-care cardiac ultrasounds, and they can be caused by trauma, malignancy, autoimmune disease, renal disease, and infection, or be idiopathic. Emergent pericardiocentesis is often indicated in the setting of cardiac arrest and in patients with hemodynamic compromise. Because it is a procedure that is uncommonly performed, operator confidence when performing this procedure based on landmarks only may be limited. The success of a pericardiocentesis is dependent on the size of the effusion and distance from the skin to pericardial fluid. It is recommended to choose the needle insertion site based on the location at which the effusion lies closest to the skin to avoid vital structures. Point-of-care ultrasound allows the operator to visualize the real-time needle throughout the entire procedure. In general, continuous observation of the advancing needle tip is challenging in the subxiphoid compared to the parasternal approach. The apical technique may be preferred over the subxiphoid but still proves much more difficult than utilizing the parasternal window in many patients.

Our study results show that the approach with the highest complication rate is the subxiphoid approach. This result is not unexpected given that the needle may traverse vital structures such as the lung, liver, internal thoracic artery, left anterior descending artery, colon, and stomach.^[16] This is also the longest distance from skin to pericardial space. In Petri et al's simulated study of 150 blind pericardiocentesis, the complication rate ranged from 5% to 31% with different puncture techniques and authors recommended to perform pericardiocentesis with image guidance whenever it is possible.^[16] Ultrasound-guidance is especially important when the pericardial effusion is posterior in location and the usual approach from the subxiphoid position would not reach the pericardial space.

Osman et al^[17] evaluated a novel in-plane parasternal medial-to-lateral approach for pericardiocentesis with a high-frequency probe in patients with cardiac tamponade and reported a 100% success rate without complications. The advantages of this medial-to-lateral approach include better visualization of needle trajectory and adjacent anatomical structures (lungs and thoracic vasculature) with a high-frequency probe which can reduce complications and shorten procedural time. This differs from the traditional technique (subxiphoid and apical) using a low-frequency probe where real-time visualization of the entire needle trajectory might be challenging and can potentially result in puncture of adjacent vital organs. The potential complications include pneumothorax with an apical approach, and increased risk of injury to the liver, heart, and inferior vena cava with subxiphoid approach. Several other observational studies demonstrated that the parasternal approach was superior to the traditional subxiphoid approach.^{[18,19,20,21]} In our study, the parasternal approach had the lowest predicted complication rate with a shorter mean skin-to-pericardial fluid distance compared to the subxiphoid approach. Our study suggests that the parasternal approach is superior to the subxiphoid approach even under ultrasound guidance. We recommend choosing a parasternal approach whenever possible, since it provides the safest, superficial, and direct path to the pericardial space. However, we acknowledge that other factors may determine the site of pericardiocentesis with complete disregard to the ideal site based on imaging. If a patient is in cardiac arrest, this precludes a parasternal approach entirely due to cardiopulmonary resuscitation. In these situations, a subxiphoid approach is wholly preferred. Patients with chronic obstructive pulmonary disease with hyperinflated lungs are unlikely to have preferable parasternal or apical views, and a subxiphoid approach may be preferred in these situations.

Probe choice for emergent pericardiocentesis can differ based on the chosen approach and patient characteristics. A linear array probe may be chosen for effusion that is directly opposed to the chest wall and can be fully imaged within a few centimeters of depth. This may be ideal for a parasternal approach or an apical approach if the effusion is amenable to superficial imaging. A linear array probe allows for improved spatial resolution with fewer artifacts. Additionally, in-plane needle guidance with a linear array probe is a skill that many emergency physicians utilize on a regular basis, making this rare procedure more familiar to other common procedures (i.e., ultrasound-guided regional anesthesia and peripheral and central vascular access). Utilizing the same in-plane needle guidance technique with a phased or a curvilinear array probe does not provide the level of detail with needle guidance and is more prone to confounding artifacts. The limitation of depth with the linear probe precludes its use when the pericardial effusion is deep or if all relevant structures (i.e., the heart) cannot be imaged within the field. In these scenarios, a phased array or curvilinear probe is ideal, despite the anticipated difficulties with needle guidance.

Our study has several limitations, including its retrospective study design and also a small number of pericardial effusions, which can limit the conclusions that can be reached. Another major limitation of this study is the selection bias from the convenience sample design, since emergency department patients received point-of-care cardiac ultrasound examination only when credentialed attending physicians were on duty. The image reviewers were not blinded to the study hypothesis. Not all three views (subxiphoid, parasternal, and apical) were obtained in every point-of-care cardiac ultrasound examination. This is likely because the studies were performed by residents and attending physicians with varied ultrasound experience which reflects the real-world clinical setting. We did not review medical records to determine the clinical significance of pericardial effusion and if they underwent pericardiocentesis; therefore, the final outcomes (arrhythmia, cardiac arrest, mortality, etc.) including actual complication rates of pericardiocentesis are unknown.

Our results suggest that complication rates with pericardiocentesis will be lower via the parasternal or apical approach compared to the subxiphoid approach. The distance from skin to fluid collection is the least in both of these views. However, additional factors may determine the best approach for pericardiocentesis regardless of the ideal site found on ultrasound, and a patient-specific or individualized approach should be considered when using ultrasound guidance.

Funding: None

Ethical approval: An IRB approval was obtained for this study.

Conflicts of interest: Authors have no financial or other conflicts of interest related to this submission.

Contributors: LS proposed and wrote the first draft. All authors contributed to the design and interpretation of the study and to further drafts.