Is it Time to Replace Physical Examination with a Hand-Held Ultrasound Device?

Attempts at using physical examination (PE) go back centuries, with inspection, palpation, and percussion being the mainstay of this approach until 2 centuries ago when the stethoscope was invented and auscultation became probably the most important element of PE for patients with known or suspected cardiovascular disease (CVD). Despite its several limitations, PE is still used, sometimes as the only means, of evaluating and following patients with CVD. In this paper I shall argue for the substitution of this inaccurate and archaic approach by direct visualization of the heart using a hand-held ultrasound (HHU) device. I am not in any way suggesting the substitution of a comprehensive echocardiographic examination by an expert sonographer/echocardiographer by HHU in patients with significant CVD. Instead, I am arguing for the replacement of PE for evaluation of the heart at the point of care as well as at the bedside, simply because HHU is more accurate and provides more meaningful information.

INTRODUCTION

Value of Physical Examination (PE)

Let us start by assessing the limitations of PE. In a study of 453 residents and 88 medical students, Mangione and Nieman found that only a minority (< 50%) were able to identify 12 most commonly encountered cardiac abnormalities (including murmurs and extra sounds such as S3).[1] In another study Oddone et al., randomized 56 interns to a teaching group (using the patient simulator Harvey) versus a nonteaching group; in no instance was any of the three conditions (mitral stenosis, mitral regurgitation, or aortic regurgitation) identified more than two-third of the time even with eight sessions on Harvey. Mitral stenosis was identified less than one-fourth of the time.[2]

Ishmail and colleagues determined the agreement among four trained observers on the presence of an S3 in 81 hospitalized patients. Agreement varied between 48 and 73% among the observers with a kappa statistic that was modest at best (0.40–0.50) and poor (0.10–0.30) at worst. The rate of agreement did not appear to be affected by the time interval between measurements or by a training effect over the time of the study.[3] Similarly, Lok and Morgan used phonocardiography as the reference standard to detect S4 and S3 in 40 patients with a cardiac diagnosis and six controls. Participants were two cardiologists, one internist, three senior residents, and two junior residents. Interobserver agreements were dismal for detecting S4 (kappa = 0.05) and S3 (kappa = 0.18) with no apparent trend in the accuracy or interobserver agreement with regard to the level of observer experience.

Comparison of Hand-Held Ultrasound (HHU) to PE

Now let us compare HHU to PE in making a cardiovascular disease (CVD) diagnosis. The early comparisons were made using older devices that did not fit into the pocket. Spencer et al., were among the first to report on the comparison of such a device to PE, where 36 patients were evaluated by[4] cardiologists against HHU using standard echocardiography as the reference standard. While PE missed 59% of the findings, HHU missed only 29%.[5]

We recently completed a similar study using a newer device that fits in the pocket.[6] The reference used was a standard echocardiogram. The indications were broadly grouped under the five most common categories seen in our echocardiography laboratory: Left ventricular (LV) function in patients with chest pain, dyspnea, etc.; valve disease in patients with murmurs or known valve disease; cardiac source of embolism in patients with stroke; structural heart disease in patients with arrhythmias; and miscellaneous (congenital abnormalities, hypertrophic cardio-myopathy, diseases of ascending aorta, pericardial effusion, etc.). Since some patients were referred for more than one indication, each patient was assigned a primary and secondary (if needed) indication.

A cardiology fellow was assigned to the echocardiography laboratory in order to identify patients referred for one of the above five indications. The number of patients selected for each indication was roughly proportional to the frequency with which they were referred to the echocardiography laboratory. The fellow then identified a cardiology attending physician who examined the patient without taking any history and who was blinded to the echocardiogram results. The patient was also examined the same day by another cardiologist using HHU who also was blinded to the standard echocardiogram findings. Each attending was only told of the indication for which the standard echocardiogram had been ordered. They then completed a predesigned form that included their findings.

There were 17 cardiologists who performed the PE and were classified according to the level of clinical experience based on the number of years they had practiced as an attending physician (< 5, 5–10, and > 10). There were four cardiologists who performed HHU examination. They all had some experience in echocardiography varying from < 2 years to > 20 years.

The data were analyzed in two steps. The first was a comparison of HHU and PE for any abnormal finding that was noted on the standard echocardiogram, which was considered the reference standard and which was interpreted by trained cardiologists assigned to the echocardiography laboratory. Mild abnormalities were not considered different from normal. Abnormalities (especially in terms of valve findings) had to be substantially different (e. g., none versus moderate or mild versus severe) to be considered disparate between the two methods. The second analysis was the comparison between HHU and PE for the primary and secondary indication for the standard echocardiogram.

We recruited 250 patients (164 men and 86 women) ranging in age from 19 to 101 years (mean 61 ± 15). The mean body mass index (BMI) was 29.9 ± 6.6. A BMI of > 30 was noted in 109 (44%) patients. One hundred and seven patients had at least one finding not assessable by PE. The mean BMI of these patients was not different from the 143 patients in whom all findings were assessable on PE (29.8 vs 30.0). Similarly, the BMI of the 29 patients in whom at least one finding was not assessable by HHU was no different than the 221 in whom all findings were assessable (31.7 vs 29.7). The time taken for PE time was 5 ± 3 min, while that for HHU was 8 ± 3 min.

Of the 250 patients, 142 had an abnormal finding on standard echocardiography. Of these HHU correctly identified 117 patients (82%) and PE correctly identified only 67 (47%) with at least one abnormality seen on the standard echocardiogram. Table 1 compares the results of HHU and PE based on findings of each irrespective of the clinical indication. HHU was vastly superior to PE for both normal as well as abnormal LV and RV function. Both approaches were equivalent in determining the absence or presence of pulmonary hypertension, with neither approach performing particularly well in detecting the presence of pulmonary hypertension. Of note, unlike the standard echocardiogram, the HHU equipment does not have spectral Doppler and hence the tricuspid jet velocity used to calculate the pulmonary artery pressure cannot be measured.

Table 1

Correct diagnoses by HHU and physical examination (PE) with the standard echocardiogram as the reference

Echocardiogram finding	HHU % correct	PE %correct	% difference	P - value (95% CI)
Normal LV function (n = 196)	89%	58%	31% (23%, 39%)	< 0.0001
Abnormal LV function (n = 54)	96%	35%	61% (45%, 77%)	< 0.0001
Normal RV function (n = 203)	94%	57%	37% (30%, 45%)	< 0.0001
Abnormal RV function (n = 47)	68%	21%	47% (26%, 67%)	0.0001
Pulmonary hypertension absent (n = 191)	92%	89%	3.1% (−3%, 9.3%)	0.36
Pulmonary hypertension present (n = 59)	53%	42%	10% (−8.3%, 28.6%)	0.33
Valve disease, mild or absent (n = 199)	94%	91%	3.5% (−1.9%, 8.9%)	0.23
Valve disease, moderate or severe (n = 51)	71%	31%	39% (19%, 59%)	0.0003
Miscellaneous findingsa absent (n = 143)	77%	64%	13% (1.7%, 23.5%)	0.02
Miscellaneous findingsa present (n = 107)	47%	3%	44% (33%, 55%)	< 0.0001

Adapted from Mehta et al.[6] aMiscellaneous findings include LV, RV, and aortic dilation; LV hypertrophy, hypertrophic cardiomyopathy, atrial and ventricular septal defect, and other congenital abnormalities; and pericardial effusion. HHU = Hand-held ultrasound, CI = confidence interval, LV = left ventricular, RV = right ventricular

Where as both methods could reliably exclude substantial valve disease (moderate or severe stenosis or regurgitation), HHU was far superior in terms of correctly identifying the presence of such disease. HHU was also marginally superior to PE in excluding several miscellaneous findings, but was far superior in terms of identifying these findings when they were present, although the accuracy of HHU was also not exemplary in this situation.

Table 2 illustrates the performance of HHU and PE in the assessment of substantial (moderate or severe) valve disease. Only the three diseases that were present in more than 10 individuals in this population are listed. For all three conditions (mitral as well as tricuspid regurgitation, and aortic stenosis) the two methods were equivalent in terms of excluding disease. However, HHU was markedly superior in identifying the presence of moderate to severe mitral and tricuspid regurgitation, but not aortic stenosis.

Table 2

Accuracy of HHU and PE in moderate to severe cardiac valve disease based on standard echocardiogram

Echocardiogram findings	HHU % correct	PE %correct	% difference	P – value (95% CI)
Mitral regurgitation absent (n = 230)	99.6%	97%	2.6% (−0.2%, 5.4%)	0.07
Mitral regurgitation present (n = 230)	100%	60%	40% (14%, 66%)	0.008
Tricuspid regurgitation absent (n = 225)	97%	98%	-0.9% (−4.1%, 2.3%)	0.75
Tricuspid regurgitation present (n = 25)	88%	28%	60% (31%, 89%)	0.0007
Aortic stenosis absent (n = 234)	97.9%	96.6%	1.2% (−1.9%, 4.4%)	0.55
Aortic stenosis present (n = 16)a	93.8%	87.5%	6.3%	1.0

Adapted from Mehta et al.[6]There were only 10 patients with moderate or severe aortic regurgitation, tricuspid or mitral stenosis, or pulmonary valve disease. aSmall samples prevented the calculation of a reliable CI. HHU = Hand-held ultrasound, PE = physical examination, CI = confidence interval

Table 3 depicts the performance of PE and HHU based on the clinical indication for the original echocardiogram. HHU was vastly superior to PE for assessing LV function in the 172 patients referred to the echocardiography laboratory with chest pain or dyspnea as the primary or secondary indication. Both normal and abnormal LV function were far better assessed by HHU compared to PE. In these 172 patients, the putative cause of patient symptoms was correctly assessed in 152 patients (88%) by HHU and only 78 (45%) by PE.

Table 3

Rate of correct diagnosis by HHU and physical examination (PE) based on initial echocardiogram indication

Clinical indication	Echo parameter	HHU correct	PE correct	P - value
Chest pain/dyspnea (n = 172)	Normal LV function	106 (85%)	59 (48%)	< 0.0001
	Abnormal LV function	46 (96%)	19 (40%)	< 0.0001
Murmur (n = 50)	No or mild valve disease	38 (100%)	29 (76%)	0.004
	Moderate or severe valve disease	8 (67%)	4 (33%)	0.22
Stroke (n = 26)	No abnormality	10 (83%)	8 (67%)	0.69
	Any abnormality	8 (57%)	4 (29%)	0.29
Arrhythmia (n = 20)	No abnormality	9 (90%)	3 (30%)	0.03
	Any abnormality	8 (80%)	3 (30%)	0.18

Adapted from Mehta et al.[6] The indications included both primary and secondary indications. For any abnormality to be correct at least one disease had to be correctly detected. Valve disease included stenosis or regurgitation of any of the four cardiac valves. HHU = Hand-held ultrasound, LV = left ventricular

Moderate or severe valve disease was rarely the cause of murmur in the 50 patients with this indication. However, HHU correctly identified all the 38 patients who did not have significant valve disease compared to only 29 such patients identified by PE. In terms of finding a cardiac source of stroke almost half (12) of these 26 patients had no cardiac abnormalities on standard echocardiography. Eight of the 14 patients with stroke who had some abnormality on standard echocardiography were correctly identified by HHU compared to only four on PE. Similarly of the 20 patients with arrhythmias, cardiac structural abnormalities were found in only 10 on standard echocardiogram. Of these, eight were correctly identified by HHU compared to only three with PE. Miscellaneous indications were too few for meaningful comparison.

Among the 17 cardiologists who performed PE, four had < 5 years of experience and performed 23.5% of the examinations, three had 5–10 years of experience and performed 17.6% of the examinations, and 10 had > 10 years of experience and performed 58.8% of the examinations. Except for the presence of pulmonary hypertension, greater experience was not correlated with a more accurate diagnosis. Correct diagnosis by HHU was also not related to the echocardiography experience of the cardiologist for any of the conditions.

Effect of HHU on cost

In our study, the physicians also indicated on the form whether the patient either needed no further testing or needed to undergo one of the five following tests to further clarify the diagnosis: formal echocardiography; stress testing with imaging, cardiac magnetic resonance imaging, cardiac computed tomography including examination of the coronary arteries; or cardiac catheterization including coronary angiography. The charges assigned for downstream testing were then obtained from the hospital billing department.

Of the 142 patients with at least one abnormality on the standard echocardiogram, further testing was suggested in 128 patients (90%) after PE and 129 patients (91%) after HHU. Further testing suggested was standard echocardiography in 181 patients having HHU and 207 patients undergoing PE. The number of patients that would have been referred for cardiac catheterization was similar for HHU and PE (7 versus 6). There was a marginal increase in charges after HHU examination. In comparison, of the 108 patients without any abnormalities on standard echocardiography, further testing was suggested in 89 (82%) undergoing PE versus only 60 (56%) undergoing HHU. There were appreciable savings in charges because of the reduction in the number of downstream testing suggested after HHU. Most of the cost savings were achieved by a more accurate assessment of LV function by HHU. Further cost savings could be achieved if bedside HHU precludes further more involved testing and resulting in reduction in hospital stay and earlier discharge.

Similar results have been reported by Cardim and colleagues in 189 outpatients who underwent PE followed by HHU examination.[7] As a result of improvement in diagnosis, HHU resulted in a lower number of patients sent to the echocardiography laboratory, thus saving cost. Greaves and colleagues examined 157 consecutive inpatients referred for echocardiography with HHU and concluded that if only those with abnormal HHU had undergone standard echocardiography, it would have reduced the number of standard echocardiograms by close to 30%, resulting in considerable savings.[8]

Can physicians be trained in HHU?

Obviously, it is not only cardiologists with training in echocardiography who perform PE on patients with CVD. Thus, it is important to know whether non-cardiologists can be trained in HHU and how long would it take to train them in basic views and assessment of common conditions. HHU does not have all the functionality of standard echocardiography, such as spectral Doppler, strain imaging, myocardial perfusion imaging, etc. There are several studies addressing this issue.

In one of the earlier studies on this subject, Kimura and colleagues trained 13 medical residents on HHU by giving a 1 h lecture on PE for LV dysfunction followed by a 1 h HHU training session, which included ‘hands on’ examinations on five volunteers. Six weeks later they were asked to examine patients and controls by both PE and HHU. HHU training improved the PE, but more importantly decreased the error of PE in 10 of 13 patients. Adequate quality images were obtained in > 80% of patients.[9]

In a more recent study by Panoulas and colleagues, five final-year medical students and three residents without prior echocardiographic experience participated in a standardized 2 h bedside tutorial on HHU. Subsequently, they assessed cardiology patients using history, PE, electrocardiogram, and HHU. A total of 122 examinations were performed. The diagnostic accuracy after history, PE, and electrocardiogram interpretation was 0.49± 0.22 (maximum = 1), whereas, the addition of HHU significantly increased its value to 0.75 ± 0.28. When assessing LV dysfunction by means of history and PE, specificity was 85% and sensitivity only 26%, whereas, after including findings from HHU, these figures rose to 94 and 74%, respectively.[10]

In a report by Alexander and coworkers, 20 medical residents participated in a 3 h HHU training session followed by examination of 534 patients undergoing standard echocardiography. They were found to perform moderately well (kappa = 0.31-0.51) compared to standard echocardiography in terms of detecting LV dysfunction, significant mitral regurgitation, aortic valve disease, and pericardial effusion.[11]

Kobal et al., compared CVD detection by medical students using HHU to experienced cardiologists with standard echocardiography as the reference standard. The students received 18 h of HHU training over 4 weeks including 14 h of ‘hands on’ experience (performance and interpretation of studies). Of the 239 abnormal CVD findings on standard echocardiography, the students detected 75%, while the cardiologists detected only 45%. The specificities of the abnormal findings were also higher for the students (87 versus 76%). For the most serious abnormalities, students performed much better than the cardiologists (96 versus 68%).[12]

Hellmann and colleagues reported on the rate at which 30 residents learned to use HHU. They received formal training in HHU consisting of 15–30 min of didactic instruction followed by ongoing one-on-one instruction in performing HHU from a sonographer. The residents’ HHU results were compared with standard echocardiography. Linear regression models showed that the residents’ overall technical proficiency skills improved at the rate of 0.79 points (0–3 scale) per 10 scans completed. Interpretation accuracy improved at a rate of 1.01 points per 10 scans as measured by an interpretation accuracy index (0–3 scale).[13] Thus, residents learned HHU at a reasonable rate.

Obstacles to use of HHU

There may be several reasons why HHU has not been more readily adopted by cardiologists in their practice. First may be the reluctance on the part of most physicians to obtain additional, albeit minimal, training in the use of HHU. Second may be the perception that a HHU examination takes considerable more time than PE, thus making it unfeasible in a busy practice. Third, there is no financial or other incentive, for it takes more time without additional compensation. Fourth, at least among some cardiologists, there may be the concern that it will reduce the need for a standard echocardiogram that may adversely affect their income in the current fee-for-service setting. Fifth is the opposite concern that based on HHU spurious echocardiograms will be ordered, increasing overall cost. Sixth is the argument that HHU is not as accurate as the standard echocardiogram.[1415161718]

Since HHU is so much more superior to PE in terms of making a diagnosis and potentially reducing need for unnecessary tests, these arguments have very little rational basis no matter how reasonable they might appear. Physicians constantly receive continuing medical education and a 2 day course on the use of HHU complimented with readily available supplementary online training could easily be implemented. The perception that HHU takes more time is true, but only by a small margin in absolute terms. For the additional time taken and given that HHU use can potentially reduce cost to the system, its use should be incentivized.

HHU could reduce referrals for the standard echocardiogram, at least among cardiologists and even among other physicians as they become accustomed to its use. In a fee-for-service environment this may be a disincentive, but in a capitated one this may contribute to lowering healthcare costs. It is also true that HHU is not as robust as a standard echocardiogram for confirming or excluding cardiac findings, but that is not its purpose. Its purpose is to replace a hundreds of years old technology (the stethoscope) with a new one.

Summary

HHU is easy to use, especially for cardiologists. It can help enormously at point-of-care and at the bedside in terms of making a diagnosis and reducing downstream testing and cost. It can be taught to other physicians, thus increasing its utilization in medicine. It has been introduced in the PE section curriculum of many medical schools, including ours.

These devices are also being used more commonly to assist vascular access[19] and removal of fluid from the thorax and pericardium.[20] They are being placed in critical care units or emergency rooms to assist in the evaluation of hemodynamically compromised patients in order to rule out severe LV dysfunction, pericardial effusion, pulmonary embolism, etc.[2122232425] Guidelines have been developed by professional societies for the use of focused ultrasound examinations in these settings.[141518] HHU has also been used as a screening and diagnostic tool in locations without access to the standard echocardiogram.[26272829] Images have been interpreted either on site or remotely through some form of telemedicine or web-based access to uploaded images.[282930] All around us technology is moving at a breathtaking pace and influencing our daily lives in ways unimaginable just a few years ago. Yet we walk around with the stethoscope in our pockets or around our necks! It is time that we put aside this 200-year-old technology and embrace the modern world. Otherwise we will look like dinosaurs!