Our study shows 60 out of 68 patients (88%) received correct disease-specific treatment using the above diagnostic strategy with agreement in ED diagnosis and final discharge diagnosis. The measure of agreement Kappa is 0.805 (p .000) where the level of agreement is statistically significant. Patients who expired were excluded for evaluation of diagnostic strategy as many of these expired within 48 h of presentation and were not able to complete their in-hospital evaluation. Most of the discrepancies occurred in patients with acute RV failure if it was due to an acute event like pulmonary embolism or acute exacerbation of chronic pulmonary artery hypertension and in patients having additional component of COPD in ADHF patients. Two studies published in 2014, similar to the present study, assessed the impact of multi-organ POCUS, in addition to history and physical examination, on the accuracy of treating the patient. In a randomized controlled trial (RCT) where patients were randomly assigned to initial assessment with and without point-of-care ultrasonography (POCUS), Pirozzi et al. found that the rate of discordance between initial and final diagnosis was 5% in the POCUS group compared to 50% in the control group . Lauresen et al. found a proportion of correct presumptive diagnosis in the POCUS group of 88% compared to 63.7% in the control group, a significant difference .
The most common diagnosis for dyspnea in this study was ADHF (43%). Sensitivity and specificity of the diagnostic strategy used in this study to identify ADHF was 97.3 and 93.3%, respectively. With regard to the test performance characteristics of POCUS as a stand-alone test for ADHF, Kajimoto et al. found a sensitivity and specificity of 94 and 91%  and Russell et al. reported sensitivity and specificity of 83 and 83% [9, 10]. As opposed to the more comprehensive and time-consuming echocardiography protocols used by other investigators [19, 20], the echocardiography component of our study protocol simply focused on ejection fraction by gross visual estimation [21, 22] (an adopted method by ACEP for emergency cardiac USG to assess global LV systolic function), presence or absence of pericardial effusion, and right ventricular enlargement while in previous studies, they evaluated diastolic function and Doppler evaluation of the heart. To diagnose diastolic heart failure, we have taken a set of parameters like LVH, LA enlargement, bilateral B-lines on lung USG, and dilated IVC along with suggestive clinical signs. Similarly, the lung examination consisted of assessment of ten zones bilaterally. We have attained similar sensitivity and specificity using this abbreviated protocol. Not only is such an abbreviated protocol feasible during initial resuscitation of the sickest dyspneic patients, but it is likely to be more generalizable to non-expert sonographers across all settings.
To date, there have been just a few studies evaluating a multi-organ POCUS protocol similar to our study–combining abbreviated echocardiography, lung USG, and IVC assessment in the setting of undifferentiated dyspnea. In addition, we included renal USG in the present study to assess kidney size and echotexture. The majority of these studies focused strictly on diagnosis of ADHF [9, 10] while the present study went beyond just ADHF diagnosis. Among the study subjects, dyspnea was attributed to ADHF in 43%, COPD exacerbation in 4%, ARDS in 7%, acute pneumonia in 4%, massive pleural effusion in 3%, acute pulmonary embolism in 7%, and AKI with volume overload and metabolic acidosis in 4%. Patients having both cardiac and non-cardiac cause of dyspnea who accounted for 28% of the study population mostly had a diagnosis of ischemic cardiomyopathy (ICMP) with CKD, COPD/ILD with RV failure, and ADHF with pneumonia. In previous studies by Pirrozi et al. and Laursen et al., acute exacerbation of COPD and acute pneumonia constituted 31.3 and 30% of their study population [11, 18]. In the PRIDE study, COPD and pneumonia constituted 25 and 10.7%, respectively . This difference can be attributed to the fact that most of the COPD and pneumonia patients were OP visits rather than ED visits at our setting and we have not enrolled known COPD patients for whom treating physician lists no other possible diagnosis.
On logistic regression analysis of, at-admission patient characteristics, IVC diameter, EF by eyeball method, and lung sliding showed independent association between cardiac and non-cardiac diagnosis. Among the clinical variables, h/o fever and cough, and jugular venous distension showed independent association. Abnormal ECG and Boston criteria were not independently helpful to differentiate between cardiac and non-cardiac diagnosis. A study by Prosen et al. showed there is significant difference in modified Boston criteria for HF score between cardiac (mean 10.9 ± 1.8) and pulmonary (4.6 ± 1.2) patients , while in our study, it showed little difference in scores between cardiac (mean 8.07 ± 1.3) and non-cardiac group (6.86 ± 1.15), questioning the reliability of score in differentiating HF from non-cardiac causes. This difference may be because we included patients with pulmonary, metabolic, and other systemic causes of dyspnea in our non-cardiac group in contrast to other studies which included only pulmonary as non-cardiac.
While it is well established that the presence of AIS is fairly sensitive for detecting ADHF , it is possible to have AIS without ADHF. B-pattern alone was not statistically significant between cardiac and non-cardiac groups as interstitial pneumonitis, pulmonary fibrosis, and ARDS will also show a similar pattern. Combined lung-cardiac-IVC USG allowed us to differentiate accurately between these groups. Pleural effusion did not add to B-profile in identifying ADHF in our study as it was in LUCUS protocol study. We faced certain diagnostic challenges in differentiation between diastolic heart failure and pulmonary pathology as both can show B pattern, i.e., to distinguish between wet and dry B-lines. The entire clinical picture helped us in reaching the diagnosis such as - a hypertensive with LVH and LA enlargement with B-pattern was in favor of diastolic heart failure and normotensive with h/o chronic respiratory disease and B-pattern with dirty appearing lungs (fragmented pleural line, subpleural abnormalities) was suggestive of a pulmonary pathology.
We included renal USG in the study as the prevalence of renal failure is high in our setting. Twelve percent of our study population had kidney size of < 9 cm. Combining renal USG to cardiopulmonary USG provided additional diagnostic data in our study population. We did not included BNP in our study as it can be elevated in the setting of CHF when an etiology other than ADHF actually accounts for the acute dyspnea and questionable economic gains and patient benefits of subjecting every patient with dyspnea to BNP assay as diagnostic uncertainty exist with mid-level BNP values [25, 26]. One of the unique features of the present study was that previous studies have included only cardiac and pulmonary causes of dyspnea, but we have also included non-cardiopulmonary causes including renal and metabolic causes fitting more into real-world scenario.
A significant proportion of our patients (28%) presenting with dyspnea had both cardiac and non-cardiac cause. There is a significant increase in morbidity in these patients compared to single cause of dyspnea with increase in time to relief of dyspnea (median 36 h) and hospital LOS (mean 11.5 days). So, it is important to have comprehensive search for all the major causes of dyspnea in every patient as significant number of them could be having multiple disease processes responsible for their symptoms.
Our study has limitations, the first being a small sample size. Patients were enrolled by a single physician trained in ultrasonography. All consecutive patients presenting with dyspnea to ED were not enrolled limited by the availability of enrolling physician. Because of the small sample size, some causes of dyspnea resulted in low recurrence, limiting the reproducibility of data relative to the ability of ultrasound in detecting them. We did not find pneumothorax and cardiac tamponade cases in our case series.
The ED physician sonographer could be influenced by a suggestive clinical presentation as the sonographer is not blinded to patient clinical findings. The primary endpoint was the diagnosis on the patient discharge summary. Although the analysis has been made by two independent physicians, a cardiologist and a pulmonologist, this criterion could be questionable because the final diagnosis was based on a body of evidence including ED diagnosis.