The main findings of this systematic review are: (i) Lung ultrasound resulted in a large proportion of diagnosis changes in the ED and ICU (34% and 44% of examinations, respectively); (ii) Lung ultrasound resulted in substantial management changes in the ED, ICU, and GW (48%, 42%, and 48% of examinations, respectively); (iii) On the ED and ICU therapy changes were most frequently non-invasive (92% and 51% respectively); (iv) Lung ultrasound methodology was heterogeneous across studies; v. Moderate to high risk of bias was present across all studies.
This study shows that bedside lung ultrasound is frequently a decisive tool in different clinical settings. This is an important finding: changes to physician behavior and subsequent modification of patient care might result in improvement in patient-centered outcomes. Moreover, even when no changes are effectuated, the confirmation of clinical impression could prevent uptake of further, costly or more invasive diagnostic and monitoring modalities. Additionally, the outcomes of the included studies were absolute (change versus no change), but ultrasound may also induce modification of prior likelihood. Resulting elimination of uncertainty may prevent delays of indicated care and, in some cases, patient harm [25, 26].
This study found that the majority of ultrasound-induced changes were classified as non-invasive as opposed to invasive. This classification is limited in evaluating true effects of changes, both in diagnosis and management, on patient-centered outcomes. For example, a non-invasive change may be to abstain from increasing furosemide dose, but another may be to start rapid fluid resuscitation. At the same time, small changes in management should not be underestimated, as, for example, optimization of volume status may result in faster liberation from mechanical ventilation [27].
Consistent with previous literature, the majority of studies integrated lung ultrasound with other thoracic ultrasound modalities. This is a reasonable approach as pathologies encountered upon thoracic ultrasound are often (patho)physiologically linked. Findings on lung ultrasound may support, modify, or moderate cardiac and caval ultrasound’s findings and vice versa. Moreover, previous research also showed that an integrated thoracic ultrasound approach performs better than its individual components [28]. Other studies have expanded bedside ultrasound beyond thorax to include transcranial and abdominal ultrasound and found even higher impact on clinical management (60% and 69%, respectively) [29, 30]. Evidently, point-of-care ultrasound modalities, combined or separately, have a substantial impact on clinical-decision making.
The current study examines the use of lung ultrasound in ED, ICU, and GW; three hospital departments where point-of-care ultrasound is very relevant. Results across these hospital settings were similar. Interestingly, the use of point-of-care ultrasound has expanded beyond hospital medicine. One study showed that the introduction of point-of-care ultrasound in general practice alters the diagnostic process and results in changes of diagnosis and management in half of patients [31]. Similarly, prehospital (and rural) studies employing a wide variety of POCUS examinations found a significant benefit that can dramatically alter disposition and treatment (50% of patients) and correlated well with in-hospital diagnostic results [32, 33].
The methodology and quality assessment tables highlight weaknesses in current lung ultrasound research. Methodological inconsistencies are frequent amongst lung ultrasound investigations, may impact findings, and limit clinical reproducibility or generalizability [8, 9, 34]. The lack of comparator for any of the studies may be intrinsic to the selected outcome, but emphasizes the need for controlled and well-designed studies to study the effect of lung ultrasound beyond clinician behavior: patient-centered and hospital level outcomes. Even excellent diagnostic tools do not necessarily lead to improved patient-centered outcomes [35]. Similarly, wrongful interpretation of ultrasound accompanied by unwarranted change of management may have undesired effects. Currently, several trials are either underway or recently published that may potentially provide higher quality evidence [36,37,38].
This is the first study that systematically and exhaustively, including a total of 13 studies and 2142 patients, describes the impact of lung ultrasound on clinical-decision making. The search strategy was extensive (including three databases) to enhance identification of relevant studies. Strict methodology was used, including inclusion and exclusion criteria to increase homogeneity across studies, a recurring issue in ultrasound literature.
A limitation to the current study is that the outcome of interest, physician behavior, is not necessarily associated with improvement of patient-centered outcomes. Assessment of the latter would require randomized or blinded studies to avoid confounding factors. Moreover, ultrasound-driven intended changes in physician behavior may not necessarily be executed. Feasibility studies are required to assess actual management effectuation. Furthermore, assessment of publication bias, e.g. by a funnel-plot, was not done. Lastly, it is possible that not all studies were identified due to the requirement of English language.