Clinicians were only moderately accurate in predicting severe complications among pediatric patients with suspected community-acquired pneumonia (CAP), researchers found.
Since there is no validated prognostic tool to identify children at high risk for severe CAP, clinicians are largely forced to rely on clinical gestalt—the physician’s intuition of the severity of a patient’s condition based on signs and symptoms, epidemiological knowledge, and patient history and physical exam findings—to inform patient management, Todd A. Florin, MD, MSCE, Division of Emergency Medicine at the Ann & Robert H. Lurie Children’s Hospital of Chicago, and colleagues explained in Pediatrics.
But how reliable is clinical gestalt? One prospective analysis found that a clinician’s gut feeling that a child had a serious illness was associated with increased likelihood of severe infection; however, findings from a second prospective study found no association between clinical gestalt and disease prognosis among kids with respiratory tract infections. In order to determine the prognostic power of clinical gestalt, Florin and colleagues conducted a prospective cohort study the Catalyzing Ambulatory Research in Pneumonia Etiology and Diagnostic Innovations in Emergency Medicine (CARPE DIEM) study to assess clinicians’ ability to predict development of severe complications among kids with suspected CAP.
“Although an association exists between clinician gestalt and disease severity in pediatric CAP, ED clinicians generally have only fair ability to differentiate those children who go on to develop severe complications from those who do not,” the study authors found, noting that clinician gestalt was most sensitive when predicted risk of complications was very low and most specific when predicted risk was very high—meanwhile, gestalt performed worst when predicting outcomes in kids with low-moderate predicted risk, a category which accounts for the majority of pediatric CAP patients.
“There is thus a need to develop evidence-based clinical decision rules to supplement clinical judgment, particularly for cases in which risk may be unclear or as newer clinicians are developing their clinical acumen,” the study authors argued.
In a commentary accompanying the study, Yasaman Fatemi, MD, and Louis M. Bell, MD, both of the Children’s Hospital of Philadelphia and the department of Pediatrics at the Perelman School of Medicine at University of Pennsylvania, Philadelphia, noted that while clinical gestalt has value, “it is important to keep in mind that it is also prone to error and biases and is not sufficient to guide medical decision-making and should be combined with evidence-based decision-making tools when possible.”
They added that future research could explore the degree to which clinician gestalt is affected by “immediate contextual factors (including practice setting and type, work environment, and tools and technologies used), patient factors (such as language proficiency and parental concern), and societal factors, such as racial bias.”
For their analysis, Florin and colleagues recruited patients who presented at the emergency department (ED) at Cincinnati Children’s Hospital Medical Center and were enrolled in CARPE DIEM from July 2013 through December 2017. Kids ages 3 months to 18 years of age with signs and symptoms of lower respiratory tract infection who received chest radiograph (CXR) for suspected CAP were enrolled—patients were excluded if they were hospitalized ≤14 days before the study ED visit, had immunocompromising or chronic conditions that might predispose them to severe or recurrent pneumonia (immunodeficiency, chronic corticosteroid use, cystic fibrosis, chronic lung disease, malignancy, sickle cell disease, congenital heart disease, tracheostomy, and neuromuscular disorders), or had a history of aspiration or aspiration pneumonia.
Trained research coordinators recorded demographic, historical, and examination findings from participants, and after examination the treating clinician was asked to record their clinical impressions, including probability of the child developing severe complications of CAP—for this analysis, severe complications were defined as respiratory failure, empyema or effusion, lung abscess or necrosis, metastatic infection, sepsis or septic shock, or death. Categorical response options for approximating suspected risk included less than 1%; 1% to 5%; 5% to 10%; 10% to 25%; 25%-50%; 50% to 75%; and 75% to 100%.
The study’s primary outcome was the development of severe disease or complications, and the secondary outcome was a composite representing increasing disease severity occurring within seven days of the study ED visit “as a 4-tiered ordinal variable: mild, moderate, severe, and very severe disease,” the authors explained.
The final study cohort included 634 patients with CXR findings suspicious for radiographic pneumonia—median patient age was 3.3 years, most were boys (54.7%), and most had race reported by parents or guardian as White (62.8%).
The study authors found that most children with suspected CAP on CXR (n=468; 73.8%) were given an predicted probability of <5% for developing severe complications. Ultimately, of 634 children with suspected CAP, 37 (5.8%) experienced severe complications.
“Overall, clinician gestalt was associated with the development of severe complications (P<0.01); however, in those who developed complications, clinicians tended to underestimate risk,” they reported. “Of the 37 children who developed severe complications, most (n= 25; 67.6%) were initially estimated to have a ≤10% chance of developing such complications.”
The study authors also found that children classified as severe by gestalt (n=25) were more likely to develop severe (n=6; 24%) or very severe disease (n=16; 64%), while those classified as mild (n=617) were less likely to develop severe (n=58; 9.4%) or very severe disease (n=5; 0.8%).
However, the predictive accuracy of gestalt crumbled for those in the moderate range: “Of the 433 children classified as moderate by gestalt, 10.6% developed only mild disease, whereas 53.1% developed severe or very severe disease. Of 242 children who developed severe disease, most (n= 178; 73.6%) were initially classified as moderate by gestalt. Similarly, out of the 73 children who developed very severe disease, most (n= 52; 71%) were initially classified as moderate by gestalt.”
Notably, more experienced clinicians were slightly better than newer clinicians in evaluating risk. The overall area under the receiver operating curve (AUC) for all clinicians was 0.747, but the AUC was 0.693 for patients with five years or less of experience while those with more than five-years’ experience achieved an AUC of 0.82.
Fatemi and Bell noted that Florin and colleagues “should be commended for their study of gestalt and clinical reasoning. They remind us to be cognizant of how we arrive at a working diagnosis and predict the course of illness. Additionally, by demonstrating the limitations of gestalt, they highlight the potential to augment gestalt (and lack of experience) with evidence-based decision rules. Thus, to predict the future, there is value in systematically coordinating nonanalytic (gestalt, intuition, gut feeling) and analytic (evidence-based) reasoning to improve medical decision-making.”
Florin and colleagues noted several limitations to their study, including a relatively low number of patients with severe disease or complications; a lack of insight into how gestalt was generated for each individual patient; the criteria for what constituted mild versus severe disease was not explicitly defined to the clinicians prior to the study; sicker patients likely received more aggressive treatment, potentially forestalling complications; and the study’s generalizability may be limited due to its setting in an urban pediatric tertiary care center.
-
Clinicians were only moderately accurate in predicting severe complications among pediatric patients with suspected community-acquired pneumonia (CAP), suggesting that clinician gestalt should be augmented with evidence-based clinical decision rules.
-
Clinician gestalt was least reliable when patients were at moderate risk for CAP complications, and gestalt had better prognostic value among clinicians with more than five years of experience.
John McKenna, Associate Editor, BreakingMED™
The study was funded by the National Institutes of Health and supported by
the National Institutes of Health National Institute of Allergy and Infectious Diseases (K23AI121325 to Florin and K01AI125413 to coauthor Ambroggio), the Gerber Foundation (to Florin), National Institutes of Health National Center for Research Resources, and Cincinnati Center for Clinical and Translational Science and Training (5KL2TR000078 to Florin).
The study authors and editorialists reported no disclosures.
Cat ID: 138
Topic ID: 85,138,730,138,192,152,925
