From the systematic literature search of PubMed, Embase, CINAHL, and Cochrane Central Register of Controlled Trials (CENTRAL) databases, we identified 8377 studies that were screened based on the title and abstract (Fig. 1). As a result, 66 studies were evaluated for their full text. Most studies (85%) were excluded based on their outcomes. These studies predominantly did not quantify or specify the criteria upon which clinical gestalt was based. Ultimately, 10 studies were included in this review. Additionally, by snowballing (i.e., using the reference lists of the 10 included studies), three other studies were added.

PRISMA flowchart showing the identification and screening of the included studies. PRISMA: Preferred Reporting Items for Systematic reviews and Meta-Analyses

Studies were published in the United States (31%), the Netherlands (31%), the United Kingdom (23%) and Sweden (15%) (Table 1). In four studies (31%), healthcare providers evaluated facial cues. Five studies (38%) analyzed photographs, and four studies (31%) investigated facial cues in video monitoring. Only one study included a mixed ethnicity, while the other studies either did not address this topic or included only Caucasian participants.

The definition of clinical outcome varied among the studies (Table 2). Acute illness was defined by various criteria, such as ischemia, mortality < 72 h, and unplanned ICU admission. Four studies (31%) reported a combination of adverse events as outcomes. Four studies (31%) reported that lipopolysaccharide (LPS)-induced illness in healthy volunteers was the outcome. Two studies (15%) reported cardiac ischemia as an outcome. The included studies show that various facial cues were able to identify deterioration of the patient’s clinical condition. For example, facial cues such as decreased redness, paler lips, and tired appearance were related with acute illness. Also, patterned facial expressions, including increased sadness and disgust, and decreased happiness and surprise, were observed in deteriorating ward patients. The methodologies used to quantify the accuracy of detecting acute illness varied widely, which made pooling results across studies not feasible.

The identified facial cues were categorized into two main groups: (1) facial appearance; and (2) facial expressions (Table 3, Fig. 2,3). The distinction was made because facial appearance refers to more static, structural cues, often involving a single changed cue, while facial expressions focusses on dynamic, emotional responses, like sadness or fatigue, which typically involves a combination of changes cues. This separation enables a clearer analysis of both structural and emotional cues when recognizing acute illness.

Most frequently reported facial cues for acute illness

Figure of all facial cues for acute illness

Acutely ill patients were found to have dry mucous membranes, increasingly red eyes, and lower eye glance [5, 23]. Moreover, they closed their eyes and had hanging eyelids more often [3, 5]. Naive observers rated acutely ill patients as having paler lips and skin, a more swollen face, and droopier corners of their mouths (Figs. 2,3) [5, 20]. Also, acutely ill patients exhibit increased redness around the nasal alae. One study has indicated that visible changes in circulation, such as colour changes, skin clamminess, coldness, impaired perfusion, and oedema, suggest acute illness [12].

Patterns of facial expressions can be identified in deteriorating general ward patients [20]. Facial expressions of acutely ill patients are perceived as more sad, less happy, and less surprised [21]. Patients admitted to the ICU have a decrease in the overall number of facial expressions and an increase in the diversity of facial expressions [22]. In addition, acutely ill patients appear more tired [5].