Email UpdatesThe Healthy Weight Care Assistant (HWCA) was developed to assist pediatricians in providing evidence-based obesity management for children who were at risk for developing complications of obesity. A clinician’s needs assessment was completed prior to the development of the HWCA and the results of this internal survey were used to target areas of clinician interest and gaps in knowledge surrounding pediatric obesity. We employed techniques from the field of human-computer interaction as a method for driving clinicians to use the system. The goal of this project was increasing early identification and early intervention so that we could influence weight trajectories.
The HWCA is delivered using a web-services approach and was developed using the Care Assistant framework. It is presented directly within the electronic health record (EHR) during usual clinical workflow and includes structured documentation related to childhood obesity, diagnosis suggestions, patient specific order and referral suggestions, and access to education resources surrounding obesity treatment. It was active from 2014-2016 within several Children's Hospital of Philadelphia outpatient general pediatric offices. The Care Assistant framework itself has been in clinical use since 2006 and has been constantly updated to conform with emerging informatics standards. Information on the Care Assistant Framework can be found at: http://policylab.chop.edu/blog/defining-clinical-decision-support.
Artifact Creation and Usage
Authors: Michel JJ, MD, MHS; Mui M, MD; Herman A, MeD; Khan S, MD; DeRusso P, MD
Contributors: Dean Karavite, MSI; Jeffrey Miller, MAS; Jeritt Thayer; Robert Grundmeier, MD
Developed at the Children's Hospital of Philadelphia (CHOP) by the Department of Biomedical and Healthcare Informatics (DBHi) in collaboration with the Healthy Weight Program (CHOP).
Funding: This project was indirectly funded by the American Beverage Foundation through an unrestricted grant to the Healthy Weight Program at The Children’s Hospital of Philadelphia.
Contact: Jeremy Michel (michelj@email.chop.edu) for additional information.
Use of the Healthy Weight Care Assistant is permitted but we require that people interested in using the decision support module contact the authors prior to implementation.
When a patient's chart is opened a message is sent from the EHR to the Care Asssistant framework containing relevant patient data (designated by the Care Assistant). The Care Assistant service extracts information and organizes this information for delivery to active modules as a .json object. When a module recognizes a .json object, it is triggered to display in the EHR and the .json data drives the rendering of the module. Rendering of a Care Assistant module can be done directly in the javascript-enabled webpage, but is often assisted by a dedicated python container (dockerized on an internal server). The use of a dockerized container enhances the speed and reliability of the CDS-delivery. Information processed in the python container is delivered back to the care assistant display engine, which then renders the page (drawing on all data sources). After page rending is complete a clinician can use the CDS module at any time during the office visit. Each click within the Care Assistant triggers a save event, which prevents the inadvertent loss of data. Data is saved directly in the EHR in a dedicated location of the chart that is not viewable directly. Aliases are created for each module to support the piping of information into the patient's chart as readable text. Rich text format (RTF) is used for customization of this text.
Two zip files contain the GIT projects required to instantiate the Healthy Weight Care Assistant (see Logic Files). The first hwi-assistant.zip contains the intervention, related patient handouts and resources, and software used to improve page rendering speed. The second, hwi-service.zip contains python scripts used to extract data from the Care Assistant object and prepare it for use in the HWCA module. Of note, this is a legacy project and still uses an older data extraction and conversion process (pre- SMART/FHIR). The HWCA would likely need to be updated for future systems. Additionally, it uses custom code to align insurances with preferred laboratory services and this would need to be locally adjusted. Further details relating to the instantiation of the Healthy Weight Care Assistant are provided in the accompanying documentation (see Technical Files). This includes a general description of files within each git repository.
The HWCA requires several EHR 'hooks'. Most of these are provided through the Care Assistant framework (i.e pulling the patient data, formatting the return data). However, to fully make use of the HWCA we also needed to develop an EPIC SmartSet (see technical files). Each SmartGroup in the EPIC SmartSet is linked to a single diagnosis/orderable in the HWCA. The use of the SmartSet allowed for utilization of EPIC error checking and order validation. Additional direct EHR modifications included the utilization of a CUI to hold information returned for the HWCA and the development of system SmartPhrases to support insert of information from the HWCA into clinical documentation within the patient note.
The EPIC SmartSet is available at: https://comlib.epic.com/#?query=healthy%20weight&customers=410,&tags=&type=All&pagesize=100&pageindex=0&id=694586227
Structured code that is interpretable by a computer (includes data elements, value sets, logic)
The HWCA was designed to improve recognition and standardization of care for children with overweight or obese weight status. Review of data from patients seen within our institution significant variation among in diagnosis of overweight and obesity as well as performance of co-morbidity screening evaluations for these patients. These children are at risk for a variety of health effects, including obesity-related co-morbidities. Missed diagnoses can lead to worsening severity of obesity, which is more difficult to reverse, and to delayed treatment of co-morbidities. Improved recognition and diagnosis of overweight and obesity could improve appropriate co-morbidities screening, including laboratory intervention and where needed specialty referral. Common comorbidities related to obesity include: insulin resistance and diabetes, hypertension, obstructive sleep apnea, slipped capital femoral epiphysis, non-alcoholic steatohepatitis, asthma, hyperlipidemia, Blount’s disease, pseudotumor cerebri, polycystic ovarian syndrome, not to mention the psychological effects and impact on quality of life.
The HWCA is intended for patients ages 2-18 years. Patients over 18 use BMI and not BMI%tile so these patients would not trigger the HWCA. Patients under 2 years of use weight-for-length (WFL) and would not trigger the HWCA. Additionally the educational materials, supported workup, documentation, and suggested diagnoses are targeted for this age range.
The intended access target is primary care clinicians. While some of the material is relevant for gastroenterologist who focus on childhood obesity, some of the material is too general for this population.
The HWCA is intended to be used by primary care pediatricians and other clinicians caring for children in a general pediatric practice. It is not intended for specialty care, although it could be adapted to support care of this population.
1) A visit containing both a weight and a height are required to trigger the HWCA. Visits missing either would not generate a BMI%tile in the EHR and therefore not trigger the HWCA to display.
2) The HWCA is unable to calculate blood pressure for age, height, sex percentiles (used in pediatrics to define hypertension) and therefore suggests hypertension for patients broadly based on a simple cutoff. This was a target for future development.
3) The HWCA requires that the rendering EHR/iFrame supports IE Edge. Older versions of EPIC specifically use IE Quirks Mode and this can cause the HWCA to load slowly (taking up to 4 seconds). The HWCA is not recommended for systems running IE Quirks Mode. It has also been tested with FireFox and Chrome, although not the most current versions of these browsers.
4) Materials and resources were tailored to the CHOP pediatric practices and may have localizations not relevant or discordant from other organization practices. Educational handouts are intellectual property of CHOP and while distribution is permitted, altering is not, without express permission of the CHOP Healthy Weight Program.
5) Although guidelines were used as an evidence source, many recommendations were based on expert opinion and may not reflect the minimum quality of evidence or recommendation strength required to support CDS in other institutions. Given the level of evidence, we opted to have this appear passively to clinicians instead of interrupting care.
The HWCA draws upon current practice guidelines for pediatric overweight and obesity management.
The primary recommendation source is: Barlow SE. Expert Committee Recommendations Regarding the Prevention, Assessment, and Treatment of Child and Adolescent Overweight and Obesity: Summary Report. Pediatrics. 2007;120(Supplement 4):S164-S192.
The HWCA was developed alongside a Care Pathway (algorithm) for Obesity Prevention and management: http://www.chop.edu/clinical-pathway/obesity-prevention-and-management-clinical-pathway. Additional references are available through the clinical pathway webpage.
Additional components of the module (specifically handling of vitamin D, recognition of co-morbidities, and diagnostic work up for co-morbitieties) draw upon recommendations from other sources. These range in source strength from published guidelines, to original research, to local expert consensus.
The primary recommendation source is: Barlow SE. Expert Committee Recommendations Regarding the Prevention, Assessment, and Treatment of Child and Adolescent Overweight and Obesity: Summary Report. Pediatrics. 2007;120(Supplement 4):S164-S192.
See: http://www.chop.edu/clinical-pathway/obesity-prevention-and-management-clinical-pathway for a summary of the recommendations underpinning this CDS module.
Barlow uses three recommendation strengths. As these are not independent of the quality of evidence assessment they are included here.
-
recommends with consistent evidence (CE), that is, multiple studies generally show a consistent association between the recommended behavior and either obesity risk or energy balance;
-
recommends with mixed evidence (ME), that is, some studies demonstrated evidence for weight or energy balance benefit but others did not show significant associations, or studies were few in number or small in sample size;
-
suggests, that is, studies have not examined the association of the recommendation with weight or energy balance, or studies are few, small in number, and/or without clear findings; however, the expert committee thinks that these recommendations could support the achievement of healthy weight and, if future studies disprove such an effect, then these recommendations are likely to have other benefits and are unlikely to cause harm.
Barlow uses three recommendation strength/quality of evidence assessment categories.
-
recommends with consistent evidence (CE), that is, multiple studies generally show a consistent association between the recommended behavior and either obesity risk or energy balance;
-
recommends with mixed evidence (ME), that is, some studies demonstrated evidence for weight or energy balance benefit but others did not show significant associations, or studies were few in number or small in sample size;
-
suggests, that is, studies have not examined the association of the recommendation with weight or energy balance, or studies are few, small in number, and/or without clear findings; however, the expert committee thinks that these recommendations could support the achievement of healthy weight and, if future studies disprove such an effect, then these recommendations are likely to have other benefits and are unlikely to cause harm.
These recommendations were developed by an expert committee. The recommendations represent a consensus based on the best available information. Ongoing research will eventually provide the best possible evidence for childhood obesity care, and future recommendations will reflect new knowledge.
Due in part to the quality of evidence and strength of recommendations, the HWCA was designed to appear passively in workflow. Additionally, clinicians could use as little or as much of the functionality provided as needed during each clinical encounter. Based on feedback from the Pediatric Obesity Advisory Board we did not pre-check any diagnoses (including obesity) or laboratory evaluations. Use of the documentation section and patient data drawn from the EHR drove real time display of potential diagnoses and laboratory evaluations. We considered obesity management as a longitudinal process. Each click in the system was saved in real time and documentation would persist through encounters. This was intended to support updating progress and goal and to resume the conversation (i.e. "I see last time you were you reported eating 3 servings of fruits and veggies each day, how do you feel you are doing now/"). We used the HWCA to select orders but native EHR functions to sign orders to take advantage of EHR error checking and alerting. We opted to develop our own family history section due to lack of use of the native family history section, lack of precision of the native family history section, and the ability to use the HWCA family history section to drive real time adjustments to diagnoses and laboratory evaluation suggestions.
BMI ≥ 85%tile (a.k.a. pediatric overweight status) AND either a well child visit (WCV) OR weight follow-up visit (determined by chief complaint).
Any outpatient office visits (face to face encounters) at active primary care outpatient sites.
None
The Healthy Weight Care Assistant (HWCA) is displayed within the EHR alongside the other Care Assistant modules (multiple exist including Immunizations, ADHD, Smoking Cessation). The HWCA module contains a structured documentation template, a list of diagnoses which updates in response to documentation, a list of orders which updates in response to documentation and diagnosis selection, and a set of educational handouts. Documentation is saved immediately to the chart. Diagnoses and orders are pushed through the EHR to the orders engine upon completion of the module (signified by a user click on a button), where EHR safety checks are applied to attempt to prevent errors. When the module is exited, documentation can be appended to the patient after visit summary and/or inserted in the patient note at the discretion of the clinician. Data in the care assistant is retained for future visits.
The HWCA is non-interruptive and the display of the module acts as a reminder. Internal functions within the Care Assistant include documentation templates, visualization support, CPOE support, diagnostic suggestions, and patient education support.
Data from 39,458 children with BMIs in the overweight or obese range were available for analysis, of which 2,941 had a visit at the pilot site. Recognition of obesity increased during the intervention and continued post-intervention (82% pre, 86% during, 90% post). Recognition of overweight also increased, but this trend did not continue (22% pre, 34% during, 28% post). Co-morbidity screening rates increased by 17 percentage-points during the intervention and reverted when the intervention stopped. At control sites recognition and screening rates were unchanged. When the CDS was active, there was improved weight status recognition and co-morbidity screening. The return to baseline in screenings after the intervention supports a contributory effect of the CDS, although further studies to establish this are required.