Feasibility of tele-guided patient-administered lung ultrasound in heart failure

Acute decompensated heart failure (HF) results in an excess of 1 million hospital admissions per year in the United States with approximately 25% of patients being readmitted within 30 days [1‐3]. The economic burden of HF readmissions is substantial; the average HF-related admission costs in excess of $14,000, with over half of the $39.2 billion of direct costs of HF care attributable to hospital treatment [2, 4]. Most HF admissions are the result of progressive rise in intracardiac filling pressures leading to pulmonary congestion, which is characterized by an increase in extravascular lung water [5].Abbreviations: In Abbreviations section, the term “ICD” was repeated twice. So we have remove the repeated one. Kindly check and confirm.confirmed

Unfortunately, clinical features and cardiac biomarkers have limitations in detecting early manifestations of HF, and as a result, there is considerable interest in remote monitoring technologies [6, 7] that can detect HF decompensation at an early stage. Implantable hemodynamic monitoring systems, such as the CardioMEMS device, which are placed in the distal pulmonary artery, can detect early increases in pulmonary arterial pressure and have been shown to significantly decrease HF hospitalizations by up to 37% over a 15-month period [2, 3, 7]. However, this technology is invasive, and the cost of the device and implantation exceeds $19,000, with 5-year costs exceeding $188,000 per CardioMEMS patient [8].

Lung ultrasound (LUS), an easy to use, non-invasive, and affordable technology, is an appealing alternative to current options for early detection of HF, and has shown promise in its applicability to this field. B-line artifacts, laser-like vertical hyperechoic reverberations arising from the pleural line, are indicative of extravascular lung water [9‐12]. B-lines observed with LUS correlate significantly with intracardiac filling pressures obtained by right heart catheterization and with radiographic pulmonary edema [9‐14]. Although the presence of B-lines is not specific for pulmonary congestion, the absence of B-lines is highly predictive (> 90%) for low/normal pulmonary artery occlusion pressures [11]. Multiple studies have found LUS to be significantly more predictive of the diagnosis of HF than either biomarkers or clinical assessment, in settings spanning the pre-hospital, emergency, and inpatient setting [15‐18].

In contrast to echocardiography, LUS image acquisition is simple, and windows require far less precision. Even in non-expert hands, LUS has very good accuracy in evaluating cardiogenic causes of dyspnea [9, 19‐21]. Multiple studies have demonstrated strong agreement in LUS interpretation between experts and novices [15, 20, 21]. Short training sessions allow nurses to perform and interpret LUS in the evaluation of HF with high predictive value [19].

Recently, ultra-portable, low-cost ultrasound devices with remote guidance capability have become available, opening up the possibility of various tele medicine applications. The aim of this study was to explore the feasibility of tele-guided patient-administered lung ultrasound to detect early manifestations of HF. An FDA-approved, portable, hand-held ultrasound system consisting of a probe connected to a smart phone or tablet was distributed to patients with stable HF. Patients were remotely guided through image acquisition and uploading, and the recorded clips were analyzed for image quality and diagnostic usability.

Eight stable ambulatory heart failure patients were recruited for this feasibility phase of the study (Table 1). Patient age ranged from 30 to 77 years, with mean age of 52 years. Six (75%) of the patients were male. Five (62.5%) of the patients were white, three (37.5%) were black, and two (25%) were Hispanic/Latino. Five (62.5%) of the patients had an implantable cardioverter defibrillator (ICD) in place, and ejection fraction (EF) ranged from 17% to 75%, with an average EF of 34.6%. Dry weight ranged from 52 to 160 kg, with an average dry weight of 106.3 kg.

Patients performed a cumulative total of 89 guided remote ultrasound sessions. From those 89 sessions, a total of 534 independent LUS clips were obtained and analyzed. A representative clip is shown in Fig. 2. There were no major technical issues. Two independent reviewers viewed and graded all collected ultrasound clips. 88% of all clips were interpretable per reviewer one, and 84% per reviewer two. 92.1% of LUS exams had at least 4 out of 6 interpretable clips per reviewer one, and 89.9% per reviewer two. Inter-reviewer agreement was 93%, with a Cohen’s Kappa of 0.7. More than 97% of all clips were determined to have appropriate gain and depth settings.

As mentioned above, the sensitivity for the detection of B-lines in HF patients is highest in the lateral lung zones [22, 23]. The LUS locations with the highest diagnostic accuracy for HF in this study were obtained in the mid axillary line in the third intercostal space. Although lateral LUS is harder for patients to self-perform, 75.3% of the remote ultrasound sessions produced interpretable mid axillary LUS clips bilaterally. If broadened to include the anterior axillary line, which is nearly as sensitive in the diagnosis of heart failure [22], 91.0% of the remote ultrasound sessions produced interpretable lateral clips bilaterally.

The average time to obtain all six ultrasound clips to complete a remote ultrasound session was 5.04 min. This time was calculated from the time of capture of the first clip to the time of capture of the last clip for each session, and, therefore, does not include the time to adjust the probe for the first recording. All eight patients had a shorter average time for their last three sessions than for their first three sessions, with an average time of 3.38 min for the last session, suggesting improvement in ease of use over time.

Two patients continued performing LUS without supervision, after the conclusion of their participation in the study, and transmitted their images for review. For these 13 unsupervised sessions, 100% of LUS clips were graded as interpretable.

Four of the eight patients had an implanted CardioMEMS device and had PAP recordings on the same days as their remote ultrasound sessions. While these data can only be looked at qualitatively given the small number of patients, fluctuations in PAP did not clearly track with dyspnea scores or with changes in daily weight. It is worth noting that the average dry weight for patients in this study was 106.3 kg, and that weight did not compromise the quality of the LUS clips.

Finally, in this patient group with stable HF, no abnormal B-lines were found during this study, and none of the enrolled patients were admitted to the hospital for a HF exacerbation. The highest Likert dyspnea score recorded was a three out of five, and six of the eight patients reported dyspnea scores of one throughout the study. None of the patients gained a significant amount of weight during the study. Hence, there were no clinical HF exacerbation, and no false positive B-lines were observed.

Clinical features and biomarkers have limitations in the early detection of HF, and multiple studies have found LUS—even in non-expert hands—to be significantly more predictive. Other accurate remote monitoring technologies such as the CardioMEMS device are expensive and invasive. Thus, LUS, an easy to use, non-invasive, and affordable technology, is an appealing and promising alternative.

This small study addresses the feasibility of patient-performed remote LUS, with image transmission for clinician interpretation, for the early detection of HF. While the sample size is small, we found that patient-performed remote LUS is feasible, and that it can result in a high percentage of interpretable clips that can be used to identify increasing extravascular lung water—an early harbinger of a HF exacerbation.

Very few studies have evaluated the use of patient-performed ultrasound in general. Only one recent study [24] was published looking specifically at patient-performed LUS in HF patients. In this study, participants were shown a 15 min training video on a four-point LUS self-exam and then asked to perform the same exam without assistance. While this study also resulted in a high degree of interpretable clips, the four-point exam did not include the mid-axillary line, which has the highest diagnostic accuracy for HF. In addition, each participant only performed the LUS once, in the emergency department of a hospital, immediately after watching a training video. This does not accurately simulate use of LUS at home on an on-going basis, as it would need to be used clinically to detect HF in outpatients.

Our study demonstrates that patient-performed remote LUS is not only feasible, it is also relatively quick to perform, with an average ultrasound time of 5.04 min. While this time does not include the patient set up prior to obtaining clips, that set up time is relatively minimal and can be done before a provider joins the video call. A medical video visit of 15 to 20 min is comparable in length to other clinic visits, and it would be a reasonable replacement for an in-person HF follow-up appointment.

The average dry weight for patients in this study was 106.3 kg, and yet that did not compromise the quality of the LUS clips. This is an important finding, given that many patients with HF are overweight.

Finally, the two patients who continued performing LUS without supervision after the conclusion of their study participation transmitted clips that were 100% interpretable. This suggests that patients may even be able to perform LUS without the need for provider presence after initial training and supervision. This would potentially enable longer term, ongoing remote monitoring of HF patients with many fewer clinic visits.

While these initial findings are very encouraging in terms of the feasibility of tele-guided patient-performed LUS, no HF decompensations were observed amongst these first 8 patients. To adequately determine whether the technology can be used to detect early decompensations, more patients need to be included in the study and the study period needs to be extended to witness decompensations.

In addition, the results of our study may not be generalizable to the typical heart failure population. Our patients were technologically savvy and relatively young, and may not be representative of the general heart failure population. More elderly patients with less experience with technology may not be suitable for this technology. However, this statement is not generalizable, as more patients of advanced age are becoming familiar with the use of smartphones. The study will need to be expanded to a much larger number to better define the group of heart failure patients suitable for this technology. Other issues that need to be investigated in future studies include cost effectiveness and the place of this new technology in the current clinical management landscape.

Patient-performed remote LUS in HF patients is feasible, with over 90% of exams being diagnostically valid, with at least 4 out of 6 interpretable clips. No HF exacerbations were observed in this study. Thus, to determine whether remote LUS can be used to detect early HF decompensations, the study period needs to be extended and more patients need to be included.