Introduction

Diaphragmatic dysfunction (DD) has a relatively high incidence in critically ill patients [1, 2] as a result both of disuse/atrophy during mechanical ventilation (ventilation induced diaphragmatic dysfunction, VIDD) [3] and mechanical insults such as cardiac or upper abdominal surgery [47].

In the last decade, research focused mainly on causes and mechanisms underlying dysfunction and atrophy of respiratory muscles in the critically ill, but there is still a lack of tools to monitor diaphragm activity at the bedside. Methods to assess diaphragmatic function often have low sensitivity or specificity, as in the case of chest X-rays, or are invasive and difficult to obtain at the bedside, as in the case of the gold standard twitch magnetic phrenic nerve stimulation or measurement of transdiaphragmatic pressure with esophageal and gastric balloons [8]. Diaphragmatic ultrasound (DU) in a critical care setting may be of great utility for this purpose. It is noninvasive, easily available, and allows repeated measurements.

There are two acoustic windows to explore the diaphragm. Briefly:

  1. 1.

    At the zone of apposition, between the 8th and 10th intercostal space in the mid-axillary or antero-axillary line, 0.5–2 cm below the costophrenic sinus. To obtain adequate images of diaphragmatic thickness, a linear high-frequency probe (≥10 MHz) is mandatory. At a depth of 1.5–3 cm, two parallel echogenic layers can be easily identified: the nearest line is the parietal pleura, the deeper one is the peritoneum. The diaphragm is the less echogenic structure in between these two lines (Fig. 1a). This approach is utilized to assess thickness of the diaphragm and thickening with inspiration, usually in M-mode (Fig. 1b). In healthy, spontaneously breathing subjects the normal thickness of the diaphragm at the zone of apposition is 1.7 ± 0.2 mm while relaxing, increasing to 4.5 ± 0.9 mm when breath holding at total lung capacity (TLC) [9].

    Fig. 1
    figure 1

    Diaphragm ultrasonography (DU) at the zone of apposition in a B-mode, b M-mode. 1 Thickness at end expiration, 2 thickness at end inspiration. DU, right subcostal in c B-mode, d M-mode

  2. 2.

    In the subcostal area, between the mid-clavicular and anterior axillary lines, using liver or spleen as acoustic windows. Either a cardiac or abdominal probe (2–5 MHz) can be used. Diaphragm is identified as a hyperechoic line (produced by the pleura tightly adherent to the muscle) that approaches the probe during inspiration (Fig. 1c). The inspiratory excursion can be easily measured in M-mode (Fig. 1d). In healthy subject during quiet spontaneous breathing, diaphragm inspiratory excursion was found to be 1.34 ± 0.18 cm [10]. A negative inspiratory excursion indicates paradoxical diaphragmatic movement and is associated with diaphragmatic paralysis and use of accessory muscles [11].

For a more accurate description of DU technique, we refer the reader to the related reviews [12, 13].

Ultrasound criteria for evaluation of normal and dysfunctioning/paralyzed diaphragm have been published [10, 11], but routine evaluation of diaphragm excursion and thickness is still poorly applied in daily practice.

We systematically reviewed the current literature about the use of DU in critically ill patients. The purpose of this systematic review is to answer the following question: is DU a useful and accurate method to assess DD in critically ill patients?

Methods

Two independent investigators performed an extensive search in Pubmed, Cochrane Database of Systematic Reviews, Embase, Scopus, and Google Scholar Databases, without language restrictions. References of all retrieved articles were scanned for additional relevant manuscripts.

The research string was “diaphragm*[tiab] AND (ultrasonography[tiab] OR ultrasound[tiab] OR echography[tiab])”. The research string was developed to have the widest possible sensitivity, while the specificity was guaranteed by human scanning of retrieved results as follows: one reviewer (SB) examined the titles and abstracts resulting from the electronic search to exclude articles that were obviously irrelevant. Two independent reviewers (MZ and MG) examined the full text of the remaining studies. A third reviewer (SB) was employed to make the final decision when it could not be achieved.

Studies meeting the following criteria were applied: human original studies published in peer-reviewed journals; employed prospective or retrospective design; reported the use of DU as a monitoring/diagnostic tool; enrolled patients admitted to intensive care units (ICU) for any reason. We included both adult and pediatric studies and then discussed the results separately.

Case reports, reviews, editorials, and studies available only as abstracts were excluded. Furthermore, we excluded studies performed in settings other than critical care (i.e., patients ventilated for elective surgery).

Extracted data included first author, year of publication, study design, population size, ultrasound technique used to measure diaphragmatic function (i.e., thickening or excursion, B-mode or M-mode), alternative technique to assess diaphragmatic function, main results.

In a second phase, we added a search of relevant abstracts from the last 3 years to include, as supplementary material, a list of potential relevant issues for the near future (Supplementary file 1).

This study was conducted and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. At a first screening there was no randomized controlled trial to include; therefore, usual quality assessment tools (i.e., Jadad scale) were not applicable.

Therefore, we used the QUADAS-2 tool for the quality assessment of diagnostic accuracy studies. The QUADAS-2 has the advantage of being easily fitted for observational studies investigating diagnostic/monitoring tools, assessing the risk of bias and applicability concerns in four domains: patient selection, index test, reference standard, flow, and timing [14].

The review was registered in PROSPERO International Prospective Register of Systematic Reviews (Registration Number: CRD42016036387).

Results

Twenty studies which included a total of 875 patients were finally selected [1534]. The study selection process, updated on 31 March 2016, is shown in Supplementary file 2. All included studies were published in peer-reviewed journals. No randomized controlled trials were found. All the included studies were observational, with three case/control studies. The results of quality assessment with QUADAS-2 are reported in Supplementary file 3.

Three studies [21, 24, 25] were conducted on pediatric patients, 17 on adult patients.

To assess DD, 11 studies [1519, 26, 27, 3033] measured diaphragmatic thickness, seven of them [15, 19, 26, 27, 30, 31, 33] assessing diaphragmatic contractility as thickening fraction (percentage change in diaphragm thickness with respiratory movement). Five studies [20, 21, 2830] measured respiratory excursion of the diaphragm in M-mode, five studies [2325, 29, 34] measured diaphragm excursion in B-mode, and two studies [22, 34] measured liver/spleen displacement as a surrogate for diaphragmatic excursion.

Ten studies compared ultrasound with other methods: two fluoroscopy [24, 25], four transdiaphragmatic pressure [19, 23, 26, 30], four rapid shallow breathing index (RSBI) [15, 20, 22, 33]. Table 1 summarizes the characteristics of the 20 studies selected.

Table 1 Summary of selected studies

In the selected studies, usefulness and accuracy of DU were investigated in four main settings:

To diagnose dysfunction or paralysis in critically ill patients: six studies reported the use of DU as a clinical monitoring tool to detect diaphragm dysfunction in critically ill patients. The results are summarized in Table 2.

Table 2 Summary of studies reporting DU to diagnose diaphragmatic dysfunction in the critically ill

To predict weaning success/failure from mechanical ventilation: four studies aimed to investigate the accuracy of DU in predicting extubation success or failure, two measuring excursion [20, 22] and two measuring thickening fraction [15, 33]. The results are shown in Table 3.

Table 3 Summary of studies assessing the performance of DU in predicting weaning outcome

To assess the performance of DU measurements as indexes of respiratory effort in mechanically ventilated patients: four studies assessed the accuracy of DU to assess the diaphragm workload during spontaneous or assisted breathing, one measuring excursion [23], two measuring thickening fraction [19, 26], and one measuring both [30]. The results are presented in Table 4.

Table 4 Summary of studies evaluating the accuracy of DU to assess the diaphragm muscular workload

To assess the progression of atrophy in ICU mechanically ventilated patients: six studies investigated the time course of thickness of diaphragm in mechanically ventilated patients. The results are summarized in Table 5.

Table 5 Summary of studies assessing diaphragm atrophy in mechanically ventilated patients

Reproducibility

Several studies have addressed the subject of reproducibility of ultrasound to measure the diaphragmatic displacement and thickness. Intraclass correlation coefficients (ICC) ranged from 0.876 to 0.999 (intraobserver) and from 0.56 to 0.989 (interobserver). The results are summarized in Supplementary file 4.

Learning curve

Two studies describe learning curves of trainees, one in pediatrics for excursion assessment, and one in adults for thickness measurement.

In a pediatric population, a 4-h hands-on training in ultrasound was reported, focusing on the recognition of normal and abnormal diaphragmatic motion. Semiquantitative assessment of excursion (normal/dysfunction/paralyzed) carried out by a trainee had very high repeatability compared to the one performed by an expert operator skilled in ultrasound [25].

In adult patients, the training of ultrasound operators to identify the diaphragm and measure its thickness was reported to take three to five sessions lasting 10–15 min each [15].

Discussion

This systematic review has several interesting results. First, DU is feasible at the bedside and has excellent intra- and interobserver reproducibility. Second, ultrasound is accurate in investigating diaphragm dysfunction, predicting extubation success or failure, quantifying respiratory effort, and detecting atrophy in mechanically ventilated patients.

To our knowledge, this is the first review that systematically analyzes the use of ultrasound to assess DD in critically ill patients, a composite population including both medical patients, in whom DD is mainly the result of prolonged MV, and surgical patients in whom DD is often caused by acute insults such as trauma or major surgical procedures.

The definition of ventilator induced diaphragmatic dysfunction (VIDD) in the critically ill is relatively recent [3], but its frequency and relevance are strongly enhanced in several publications [1, 35]. DD is responsible for a number of pulmonary complications, including atelectasis and pneumonia, and an early diagnosis of DD (prior to extubation) is mandatory to avoid the risk of extubation failure. Demoule et al. found that DD, defined as a reduced capacity of the diaphragm to produce inspiratory pressure, is as frequent as 64 % on the first day from ICU admission. It is associated with disease severity and sepsis, and it may represent another sepsis-related organ failure. Furthermore, it is associated with a poor prognosis [1].

Despite the widespread use of ultrasound techniques in the ICUs (namely echocardiography and lung ultrasound), DU has only recently been applied in the intensive care setting. DU allows both morphologic assessment (detection of atrophy) and functional evaluation of the muscle (contractility). Furthermore, it allows repeated measurements over time, such as before and after variations in ventilator settings, or before and after the start of noninvasive ventilation.

Several studies have compared ultrasound of the diaphragm with reference methods (i.e., transdiaphragmatic pressure) in healthy subjects, finding diaphragmatic excursion and thickening fraction very effective in assessing the diaphragmatic function [36, 37].

In our systematic review, we found DU successfully applied in four different settings:

  1. 1.

    To diagnose dysfunction or paralysis in critically ill patients. DD diagnosed with ultrasound was found in 29 % of mechanically ventilated patients without history of diaphragmatic or neuromuscular disease [20]. This finding indicates that DD is probably underestimated in ICU patients.

  2. 2.

    To predict weaning success/failure from mechanical ventilation. Either diaphragm excursion or thickening fraction measurements performed during a spontaneous breathing trial in intubated patients have shown good performance as weaning indexes.

  3. 3.

    To assess respiratory effort in mechanically ventilated patients. When compared to invasive techniques such as diaphragm and esophageal time–pressure product (PTPdi and PTPes), the thickening fraction has shown significant correlation, thus emerging as a new noninvasive tool to monitor respiratory workload during assisted mechanical ventilation.

  4. 4.

    To assess the progression of atrophy in ICU mechanically ventilated patients. Measuring thickness at the zone of apposition in mechanically ventilated patients is the best tool to detect atrophy, one of the main features (even if not synonymous) of dysfunction [2].

The technique to measure diaphragm performance varied from subcostal assessment of inspiratory excursion to assessing the muscle at the zone of apposition for thickness and thickening fraction measurements. The two techniques have indeed different features.

Thickening fraction has shown the best performance to estimate respiratory muscle workload during noninvasive mechanical ventilation and to predict extubation failure or success during a spontaneous breathing trial. The reported cutoff to predict extubation success or failure ranged between 30 and 36 % during spontaneous breathing trials [15, 33]. Nevertheless, thickness and thickening fraction measurements are not always easy to perform. First, the mean thickness values are about 1.5–2 mm and therefore it needs a high frequency probe (usually a 10 MHz “vascular” probe). Second, technical difficulties with some patients (i.e., obese patients) should be expected. Third, the smallest measurable distance of most machines is 0.1 mm, which means about 5–7 % of the measurement; therefore, small operator-dependent variations could influence the measurement. Fourth, it is not always possible to assess the left hemidiaphragm [26, 31]. Finally, there is a lack of data about the learning curve to measure the thickening fraction; nevertheless, in our experience it is longer than the one to measure respiratory excursion.

On the other hand, ultrasonographic assessment of diaphragmatic excursion is relatively easy to perform. A convex cardiac or abdominal probe should be used. The probe is placed between the mid-clavicular and anterior axillary lines, in the subcostal area, and directed medially, cranially, and dorsally, so that the ultrasound beam reaches perpendicularly the posterior third of the diaphragm. The inspiratory and expiratory cranio-caudal displacement of the diaphragm respectively shortens and lengthens the probe–diaphragm distance. To measure diaphragmatic excursion, M-mode has been shown to be more reproducible than B-mode [29].

Movement is usually better appreciated on the right side, while on the left side the descending lung, bowel, and gas interposition during inspiration often hide the diaphragm.

The best cutoff to diagnose DD with diaphragmatic excursion measurements ranged from 10 to 14 mm during normal spontaneous breathing and 25 mm for maximal inspiratory effort. It should be noted that excursion as an index of diaphragmatic function should be limited to patients on spontaneous breathing. Only one study assessed both thickening of diaphragm and excursion to evaluate inspiratory muscle effort during assisted breathing and concluded that excursion should not be used to assess diaphragm contractility [30]. In fact, excursion is mainly related to the inspired volume [37], regardless of whether it depends on muscle workload or ventilator support. Therefore, to estimate the diaphragm workload during assisted breathing thickening fraction should be measured.

Limitations

This systematic review has some limitations. The existing studies are observational, and no randomized controlled trials have been published so far on the utilization of DU in critical care; furthermore, they are relatively small and heterogeneous, and this does not allow one to perform pooled data analysis. Even if excellent reproducibility has been reported in most of the studies, attention should be drawn to the fact that statistical gold standard to assess reproducibility (i.e., Bland–Altman limits of agreement) was reported only in one publication [34]. Data on learning curves for DU are lacking, especially for thickening fraction measurements.

Only three studies compare DU with transdiaphragmatic pressure, a measure of the diaphragm’s force-generating capacity. Therefore, the relationship between diaphragm thickening or inspiratory excursion and strength of the diaphragm should be further investigated. Nevertheless, clearly all the retrieved articles support DU as a useful tool for respiratory muscle monitoring in critically ill patients.

Conclusions

DU has shown to be useful and accurate in diagnosing diaphragmatic dysfunction with a cutoff of 10–14 mm for diaphragmatic excursion and 30–36 % for thickening fraction. Current literature suggests the use of DU to detect diaphragmatic dysfunction in critically ill patients, to predict extubation success or failure, to monitor respiratory workload, and to assess atrophy in patients who are mechanically ventilated. Randomized controlled studies are needed to assess if the use of DU to guide clinical decisions may influence outcomes in critically ill patients.