Tamara M. Dugan

Targeted screening for pediatric conditions with the CHICA system

OBJECTIVE The Child Health Improvement through Computer Automation (CHICA) system is a decision-support and electronic-medical-record system for pediatric health maintenance and disease management. The purpose of this study was to explore CHICAs ability to screen patients for disorders that have validated screening criteria--specifically tuberculosis (TB) and iron-deficiency anemia. DESIGN Children between 0 and 11 years were randomized by the CHICA system. In the intervention group, parents were asked about TB and iron-deficiency risk, and physicians received a tailored prompt. In the control group, no screens were performed, and the physician received a generic prompt about these disorders. RESULTS 1123 participants were randomized to the control group and 1116 participants to the intervention group. Significantly more people reported positive risk factors for iron-deficiency anemia in the intervention group (17.5% vs 3.1%, OR 6.6, 95% CI 4.5 to 9.5). In general, far fewer parents reported risk factors for TB than for iron-deficiency anemia. Again, there were significantly higher detection rates of positive risk factors in the intervention group (1.8% vs 0.8%, OR 2.3, 95% CI 1.0 to 5.0). LIMITATIONS It is possible that there may be more positive screens without improving outcomes. However, the guidelines are based on studies that have evaluated the questions the authors used as sensitive and specific, and there is no reason to believe that parents misunderstood them. CONCLUSIONS Many screening tests are risk-based, not universal, leaving physicians to determine who should have a further workup. This can be a time-consuming process. The authors demonstrated that the CHICA system performs well in assessing risk automatically for TB and iron-deficiency anemia.

Use of a Computerized Decision Aid for Developmental Surveillance and Screening: A Randomized Clinical Trial

IMPORTANCE Developmental delays and disabilities are common in children. Research has indicated that intervention during the early years of a childs life has a positive effect on cognitive development, social skills and behavior, and subsequent school performance. OBJECTIVE To determine whether a computerized clinical decision support system is an effective approach to improve standardized developmental surveillance and screening (DSS) within primary care practices. DESIGN, SETTING, AND PARTICIPANTS In this cluster randomized clinical trial performed in 4 pediatric clinics from June 1, 2010, through December 31, 2012, children younger than 66 months seen for primary care were studied. INTERVENTIONS We compared surveillance and screening practices after adding a DSS module to an existing computer decision support system. MAIN OUTCOMES AND MEASURES The rates at which children were screened for developmental delay. RESULTS Medical records were reviewed for 360 children (180 each in the intervention and control groups) to compare rates of developmental screening at the 9-, 18-, or 30-month well-child care visits. The DSS module led to a significant increase in the percentage of patients screened with a standardized screening tool (85.0% vs 24.4%, P < .001). An additional 120 records (60 each in the intervention and control groups) were reviewed to examine surveillance rates at visits outside the screening windows. The DSS module led to a significant increase in the percentage of patients whose parents were assessed for concerns about their childs development (71.7% vs 41.7%, P = .04). CONCLUSIONS AND RELEVANCE Using a computerized clinical decision support system to automate the screening of children for developmental delay significantly increased the numbers of children screened at 9, 18, and 30 months of age. It also significantly improved surveillance at other visits. Moreover, it increased the number of children who ultimately were diagnosed as having developmental delay and who were referred for timely services at an earlier age. TRIAL REGISTRATION clinicaltrials.gov Identifier: NCT01351077.

Machine Learning Techniques for Prediction of Early Childhood Obesity.

OBJECTIVES This paper aims to predict childhood obesity after age two, using only data collected prior to the second birthday by a clinical decision support system called CHICA. METHODS Analyses of six different machine learning methods: RandomTree, RandomForest, J48, ID3, Naïve Bayes, and Bayes trained on CHICA data show that an accurate, sensitive model can be created. RESULTS Of the methods analyzed, the ID3 model trained on the CHICA dataset proved the best overall performance with accuracy of 85% and sensitivity of 89%. Additionally, the ID3 model had a positive predictive value of 84% and a negative predictive value of 88%. The structure of the tree also gives insight into the strongest predictors of future obesity in children. Many of the strongest predictors seen in the ID3 modeling of the CHICA dataset have been independently validated in the literature as correlated with obesity, thereby supporting the validity of the model. CONCLUSIONS This study demonstrated that data from a production clinical decision support system can be used to build an accurate machine learning model to predict obesity in children after age two.

Unexplained Practice Variation in Primary Care Providers' Concern for Pediatric Obstructive Sleep Apnea

OBJECTIVE To examine primary care provider (PCP) screening practice for obstructive sleep apnea (OSA) and predictive factors for screening habits. A secondary objective was to describe the polysomnography completion proportion and outcome. We hypothesized that both provider and child health factors would predict PCP suspicion of OSA. METHODS A computer decision support system that automated screening for snoring was implemented in 5 urban primary care clinics in Indianapolis, Indiana. We studied 1086 snoring children aged 1 to 11 years seen by 26 PCPs. We used logistic regression to examine the association between PCP suspicion of OSA and child demographics, child health characteristics, provider characteristics, and clinic site. RESULTS PCPs suspected OSA in 20% of snoring children. Factors predicting PCP concern for OSA included clinic site (P < .01; odds ratio [OR] = 0.13), Spanish language (P < .01; OR = 0.53), provider training (P = .01; OR = 10.19), number of training years (P = .01; OR = 4.26) and child age (P < .01), with the youngest children least likely to elicit PCP concern for OSA (OR = 0.20). No patient health factors (eg, obesity) were significantly predictive. Proportions of OSA suspicion were variable between clinic sites (range, 6-28%) and between specific providers (range, 0-63%). Of children referred for polysomnography (n = 100), 61% completed the study. Of these, 67% had OSA. CONCLUSIONS Results suggest unexplained small area practice variation in PCP concern for OSA among snoring children. It is likely that many children at risk for OSA remain unidentified. An important next step is to evaluate interventions to support PCPs in evidence-based OSA identification.

Effectiveness of Computer Automation for the Diagnosis and Management of Childhood Type 2 Diabetes: A Randomized Clinical Trial

Importance Type 2 diabetes (T2D) is increasingly common in young individuals. Primary prevention and screening among children and adolescents who are at substantial risk for T2D are recommended, but implementation of T2D screening practices in the pediatric primary care setting is uncommon. Objective To determine the feasibility and effectiveness of a computerized clinical decision support system to identify pediatric patients at high risk for T2D and to coordinate screening for and diagnosis of prediabetes and T2D. Design, Setting, and Participants This cluster-randomized clinical trial included patients from 4 primary care pediatric clinics. Two clinics were randomized to the computerized clinical decision support intervention, aimed at physicians, and 2 were randomized to the control condition. Patients of interest included children, adolescents, and young adults 10 years or older. Data were collected from January 1, 2013, through December 1, 2016. Interventions Comparison of physician screening and follow-up practices after adding a T2D module to an existing computer decision support system. Main Outcomes and Measures Electronic medical record (EMR) data from patients 10 years or older were reviewed to determine the rates at which pediatric patients were identified as having a body mass index (BMI) at or above the 85th percentile and 2 or more risk factors for T2D and underwent screening for T2D. Results Medical records were reviewed for 1369 eligible children (712 boys [52.0%] and 657 girls [48.0%]; median [interquartile range] age, 12.9 [11.2-15.3]), of whom 684 were randomized to the control group and 685 to the intervention group. Of these, 663 (48.4%) had a BMI at or above the 85th percentile. Five hundred sixty-five patients (41.3%) met T2D screening criteria, with no difference between control and intervention sites. The T2D module led to a significant increase in the percentage of patients undergoing screening for T2D (89 of 283 [31.4%] vs 26 of 282 [9.2%]; adjusted odds ratio, 4.6; 95% CI, 1.5-14.7) and a greater proportion attending a scheduled follow-up appointment (45 of 153 [29.4%] vs 38 of 201 [18.9%]; adjusted odds ratio, 1.8; 95% CI, 1.5-2.2). Conclusions and Relevance Use of a computerized clinical decision support system to automate the identification and screening of pediatric patients at high risk for T2D can help overcome barriers to the screening process. The support system significantly increased screening among patients who met the American Diabetes Association criteria and adherence to follow-up appointments with primary care clinicians. Trial Registration clinicaltrials.gov Identifier: NCT01814787