Navigation Menu

Department of Pediatrics

About

Located in north Philadelphia, the department provides clinical care to the children of our underserved and disadvantaged community. The research activities of the department have direct application to improving the health and quality of life of these children, are clinically based and fall into 3 areas: The impact of social determinants on child health, infant sleep hygiene, and medical decision making.

Social Determinants of Health and Chronic Childhood Illness

Adverse childhood events(ACEs), such as family or community violence, food insecurity or housing insecurity have been shown to have a significant impact on the physical health of adults. Resilience, as manifest by support systems, mitigates, to some degree, the physical impact of ACEs in adults. Led by Ashleigh Hall, DO, Associate Professor of Clinical Pediatrics at LKSOM, studies from our department have demonstrated a similar effect in adolescents. Current research has identified an impact of ACEs and resilience on the behavior of children aged 6 - 11 years (manuscript in preparation).

 

Sleep Hygiene in Infants

Positioning of infants in sleep, where infants sleep, and how they sleep are important health issues. While it is well known that prone positioning of newborns and young infants increases the risk of sudden unexplained death, cosleeping with parents is also a risk factor. A study by Megan Heere, M.D., Professor of Clinical Pediatrics at LKSOM, noted that infants without an identifiable place to sleep were at increaed risk for cosleeping. As a follow on study, sleep hygiene education coupled with the gifting of a cardboard bassinet at the time of discharge reduced the rates of co sleeping in our patient population. Ongoing NIH funded studies, in collaboration with investigators from St. Joesph's University and Children's Hospital of Philadelphia, are examining the relationship between eating and activity and sleep duration in children.

Profile_Megan Heere

 

Medical Decision Making

In its simplest form, clinical decisions require the establishment of a diagnosis, implementation of effective therapy and prognosis of outcome. The typical way physicians are taught to make diagnoses or develop differential diagnoses is through pattern recognition. While this approach works well most of the time and is quick, it is subject to bias. Availability (over estimation of the likelihood of the first diagnosis that comes to mind), representativeness (sterotyping symptoms into a diagnosis without supporting evidence), anchoring (overestimation of the probability of the initial diagnostic impression to the point where strong non supportive information is ignored or minimized) and adjustment (overestimation of the strength of the non-supportive data to the point where the initial diagnostic impression is dismissed) biases lead to misdiagnosis and incorrect treatment (Tversky, A. and Kahneman, D. (1974). Judgment Under Uncertainty: Heuristics and Biases. Science. 185(4157), 1124-1131.)

The probabilistic approach to differential diagnosis is evidence-based and circumvents the heuristic biases described by Kahneman and Tversky. Briefly, the clinical problem is defined and evidence from the literature is used to established the catalogue of possible etiologies along with the associated probability for each clinical entity. This "prior probability" along with subsequent diagnostic testing can be used in a bayesian system to establish the most likely diagnosis. A detailed discussion of this process is available at Aronoff SC and DelVecchio MT. Differential Diagnosis in Pediatrics: A Probabilistic Approach. Hosp Pediatr (2016) 6 (8): 504–505.

 

Evidence and Problem Based Differential Diagnosis of Pediatric Clinical Problems

Finding evidence in the literature to establish diagnostic prior probabilities is difficult. Sometimes, large single center or multicenter case series of children with the clinical problem of interest can be found. Michael Del Vecchio, M.D., Professor of Pediatrics at LKSOM, has focused on the use of systematic reviews and meta-analytical rate estimations to develop probability based differential diagnoses for common and uncommon pediatric clinical problems. By providing a research opportunity for 4th year LKSOM medical students who are the first authors on these studies, Dr. Del Vecchio has created a large body of work that is widely used in pediatrics.

Profile Michael Del Vecchio

 

Derivation and Validation of Machine Learning Predictive Pediatric Models

With the development of machine learning(artificial intelligence; statistical learning) and the availability of large clinical datasets, predictive modeling for clinical medicine has become a reality. Stephen Aronoff, MD,MBA, Waldo E. Nelson and Chairman of Pediatrics and Professor, Department of Medical Education and Data Science at LKSOM has focused on modeling the outcomes of pediatric victims of trauma. Working with the Department of Pediatric Surgery at St. Christopher's Hospital for Children and with Nathan Kupperman, Director of PECARN, machine learning models to predict the probability of clinically significant injuries in children with and were developed and validated. A model to predict the risk of deep vein thromboses in pediatric trauma victims has been completed and is currently under peer review. Links to the web applications of these models are below: