Initially described more than 50 years ago, electronic health records (EHRs) are now becoming ubiquitous throughout pediatric health care settings. The confluence of increased EHR implementation and the exponential growth of digital data within them, the development of clinical informatics tools and techniques, and the growing workforce of experienced EHR users presents new opportunities to use EHRs to augment clinical discovery and improve pediatric patient care. This article reviews the basic concepts surrounding EHR-enabled research and clinical discovery, including the types and fidelity of EHR data elements, EHR data validation/corroboration, and the steps involved in analytical interrogation.
Key points
- •
The electronic health record (EHR) contains a massive amount of discrete patient data that are generated through the routine provision of patient care.
- •
EHR data can be so-called big data based on volume (total number of patients/data points), velocity (the rate at which it is generated), and/or variety.
- •
Data validation is imperative because many of the data were collected for clinical, rather than research, purposes.
- •
EHR data can be used to build large patient cohorts and/or identify patients with rare conditions, allowing pediatric researchers to overcome small samples sizes.
- •
The EHR can be used for interventional studies and prospective trials at the point of care.
Introduction
Although electronic health records (EHRs) were first described more than 50 years ago, it is only recently that EHRs have become pervasive. Notably, the Unites States saw EHR adoption triple between 2009 and 2013; adoption among children’s hospitals increased from 21% in 2008 to 59% in 2011. The growth seen over the last 5 years is likely to be progressive and sustainable as, with time, countries such as Norway, the Netherlands, New Zealand, and the United Kingdom have achieved near-universal EHR adoption. As experience with EHRs has increased, clinicians and researchers have begun to see the EHR in a different light; as a colossal database and interventional tool at the intersection of physicians, patients, and care delivery. In the past, research databases, tools, and records have been separate from clinical databases, tools, and records; EHRs blur these distinctions and merge these silos.
In parallel with increased EHR adoption, there has been the evolution of clinical informatics, the “scientific and medical field that concerns itself with the cognitive, information processing and communication tasks of medical practice, education and research, including the information science and the technology to support these tasks.” More specifically, clinical research informatics, which generates the tools and techniques to use the EHR for clinical research, has been developed within the larger field of clinical informatics.
The combination of increased EHR implementation, the exponential growth of digital data available within the EHR, and the development of clinical research informatics tools and techniques present a unique, previously unavailable opportunity to enhance clinical discovery and improve patient care. This phenomenon has particular relevance in children’s health care. Pediatric research often involves small samples sizes ; by exploiting the EHR data sets across 1 or more institutions, investigators have access to a vast number of patients and can mitigate concerns about statistical power and significance. In addition, when faced with clinical decisions, pediatricians are often forced to rely on studies and trials that have been performed in adults; trials are often not repeated in children because of cost, safety, and redundancy issues. EHR data can be used to retrospectively assess the impact of clinical decisions, making it possible to determine the efficacy of a certain intervention or the safety of a particular medication. In addition, when pediatric trials are implemented, they are often hampered by slow recruitment and erratic referrals. The EHR can be used to facilitate recruitment for trials across institutions and improve the efficacy of trial procedures. It is within this context that this article presents an overview of EHR-enabled clinical discovery, focusing on concepts and studies relevant to pediatrics.
Introduction
Although electronic health records (EHRs) were first described more than 50 years ago, it is only recently that EHRs have become pervasive. Notably, the Unites States saw EHR adoption triple between 2009 and 2013; adoption among children’s hospitals increased from 21% in 2008 to 59% in 2011. The growth seen over the last 5 years is likely to be progressive and sustainable as, with time, countries such as Norway, the Netherlands, New Zealand, and the United Kingdom have achieved near-universal EHR adoption. As experience with EHRs has increased, clinicians and researchers have begun to see the EHR in a different light; as a colossal database and interventional tool at the intersection of physicians, patients, and care delivery. In the past, research databases, tools, and records have been separate from clinical databases, tools, and records; EHRs blur these distinctions and merge these silos.
In parallel with increased EHR adoption, there has been the evolution of clinical informatics, the “scientific and medical field that concerns itself with the cognitive, information processing and communication tasks of medical practice, education and research, including the information science and the technology to support these tasks.” More specifically, clinical research informatics, which generates the tools and techniques to use the EHR for clinical research, has been developed within the larger field of clinical informatics.
The combination of increased EHR implementation, the exponential growth of digital data available within the EHR, and the development of clinical research informatics tools and techniques present a unique, previously unavailable opportunity to enhance clinical discovery and improve patient care. This phenomenon has particular relevance in children’s health care. Pediatric research often involves small samples sizes ; by exploiting the EHR data sets across 1 or more institutions, investigators have access to a vast number of patients and can mitigate concerns about statistical power and significance. In addition, when faced with clinical decisions, pediatricians are often forced to rely on studies and trials that have been performed in adults; trials are often not repeated in children because of cost, safety, and redundancy issues. EHR data can be used to retrospectively assess the impact of clinical decisions, making it possible to determine the efficacy of a certain intervention or the safety of a particular medication. In addition, when pediatric trials are implemented, they are often hampered by slow recruitment and erratic referrals. The EHR can be used to facilitate recruitment for trials across institutions and improve the efficacy of trial procedures. It is within this context that this article presents an overview of EHR-enabled clinical discovery, focusing on concepts and studies relevant to pediatrics.
The electronic health record data set
The EHR data set is immense; the scale is on par with many of the big data disciplines such as genomics and proteomics. Across an entire children’s hospital, clinical care generates hundreds of thousands of data points per day and tens of millions of data points annually; data generated from ambulatory care and the narrative data contained within clinical notes add substantially more information. However, although the volume of data is alluring, some elements are easier to extract, some have higher fidelity, and some require validation. Thus, an understanding of the types and quality of data available is essential to using EHRs for clinical discovery.
Electronic Health Record Data Elements
The EHR contains clinical data generated through the routine provision of care: physician orders, test results, vital signs, demographics, progress notes, medication administration data, and so forth. Depending on the institution, it may include financial, operational, and technical information. Some data are discrete (vital signs), some are textual (progress notes), and some are scanned (medical records from outlying facilities). Although some elements conform to national ontology standards (eg, Common Procedural Terminology [CPT]; International Classification of Diseases 9, Clinical Modification [ICD-9-CM/ICD-10-CM]; Systematized Nomenclature for Medicine [SNO-MED], Unified Medical Language System [UMLS]), many are locally defined. Table 1 shows typical types of data present in most EHRs and an assessment of their quality.
| Type of Data | Relative Quality | Comments |
|---|---|---|
| Demographic (eg, age, gender) | Very high | EHRs overestimate the number/percentage of patients who are alive because of limited processes to update patients’ statuses unless they expire in the hospital |
| Laboratory results (eg, LOINC codes) | Very high | Data collected for clinical and not research purposes so sometimes a laboratory test was never ordered that would have been ideal; a laboratory result that does not exist is not the same as a negative laboratory result |
| Prescriptions/medications ordered (eg, RX-Norm codes) | Very high | In some cases, up to 31% of prescriptions written are not filled |
| Vital signs | High | Data collected for clinical and not research purposes |
| Test orders | High | Most EHRs have provider order entry; orders can be used as a surrogate to understand provider thought process, intent, and sometime diagnoses even if the test is never completed or results are unavailable |
| Diagnoses (eg, ICD-9/ICD-10 codes) | Medium | Highly variable; probably higher sensitivity, but lower specificity for rarer, more serious diseases |
| Family history, social history, past medical and surgical history (eg, ICD-9/ICD-10; CPT codes) | Low | Often missing; high if data collection is mandated and standardized, such as smoking status for ages 13 y and older |
| Other | Unknown | Many other data elements exist in EHRs; need to understand qualitatively and quantitatively EHR variables that you want to use |
Electronic Health Record Data Validation/Corroboration
Although validating data is always important, it is especially critical when analyzing large EHR data sets in which the data may have been collected for clinical rather than research purposes. There are 2 main methodological approaches to corroborate potential findings. One common approach is internal validation, which typically involves a detailed, manual evaluation of a random subset (sometimes only 1% or less) of the large data set to determine fidelity. The other primary approach is external validation. Here findings are corroborated by other studies or data from a different data source; findings that are consistent with previous studies or alternate data sources are more likely to be correct. Of note, if the finding of interest is a relative difference between groups whose data come from the same source, validation may be less critical as long as obvious biases are not apparent between the groups. In addition, because larger data sets make it statistically easier to find statistically significant results that may not be clinically significant (or even clinically plausible), a biological plausibility hypothesis should exist for findings and hypotheses should be developed before data analysis begins.
Electronic health record–enabled research methodologies and examples of analytical approaches
The types of studies that can be performed using EHR data typically conform to the fairly standard methodologies used with other types of data. A summary of these approaches and their advantages/disadvantages in clinical research informatics is shown in Table 2 . Detailed in Table 3 is a comprehensive list, by study type, of EHR-enabled pediatric studies as of publication.
| Research Methodology | Advantages | Disadvantages | Comments |
|---|---|---|---|
| Retrospective cohort/case-control study |
|
| Population level, deidentified data do not need IRB approval |
| Before-after technology intervention | For large implementations (such as implementing e-prescribing in a practice or health care system) no other methodology may be possible | Could be unrecognized temporal confounders | — |
| Randomized cluster control trial | Reduces individual bias or cross-contamination from patient-based or provider-based randomization | Depending on EHR, may be technically difficult to implement | A cluster would be a group of providers or single site |
| Title of Article | Investigators |
|---|---|
| Retrospective, Observational Studies | |
| Diabetes Mellitus Screening in Pediatric Primary Care | Anand et al, 2006 |
| The Natural History of Weight Percentile Changes in the First Year of Life | Bennett et al, 2014 |
| Trends in the Diagnosis of Overweight and Obesity in Children and Adolescents: 1999–2007 | Benson et al, 2009 |
| Screening for Obesity-related Complications Among Obese Children and Adolescents: 1999–2008 | Benson et al, 2011 |
| Development of Heart and Respiratory Rate Percentile Curves for Hospitalized Children | Bonafide et al, 2013 |
| Low-pressure Valves in Hydrocephalic Children: A Retrospective Analysis | Breimer et al, 2012 |
| Association Between Maintenance Fluid Tonicity and Hospital-acquired Hyponatremia | Carandang et al, 2013 |
| Heart Rates in Hospitalized Children by Age and Body Temperature | Daymont et al, 2015 |
| Evaluation of the Quality of Antenatal Care Using Electronic Health Record Information in Family Medicine Clinics of Mexico City | Doubova et al, 2014 |
| Identifying Factors Predicting Immunization Delay for Children Followed in an Urban Primary Care Network Using an Electronic Health Record | Fiks et al, 2006 |
| Impact of Immunization at Sick Visits on Well-child Care | Fiks et al, 2008 |
| Association of Late-preterm Birth with Asthma in Young Children: Practice-based Study | Goyal et al, 2011 |
| Underdiagnosis of Hypertension in Children and Adolescents | Hansen et al, 2007 |
| Increased Prevalence of Eosinophilic Gastrointestinal Disorders in Pediatric PTEN Hamartoma Tumor Syndromes (PHTS) | Henderson et al, 2014 |
| Real-time Forecasting of Pediatric Intensive Care Unit Length of Stay Using Computerized Provider Orders | Levin et al, 2012 |
| Comparison of New Modeling Methods for Postnatal Weight in ELBW Infants Using Prenatal and Postnatal Data | Porcelli & Rosenbloom, 2014 |
| Specialized Pediatric Growth Charts for Electronic Health Record Systems: The Example of Down Syndrome | Rosenbloom et al, 2010 |
| AKI in Hospitalized Children: Epidemiology and Clinical Associations in a National Cohort | Sutherland et al, 2013 |
| AKI in Hospitalized Children: Comparing the pRIFLE, AKIN, and KDIGO Definitions | Sutherland et al, 2015 |
| Retinopathy of Prematurity in English Neonatal Units: A National Population-based Analysis Using NHS Operational Data | Wong et al, 2013 |
| Interventional Quality-improvement Studies and Trials | |
| Computerized Physician Order Entry with Decision Support Decreases Blood Transfusions in Children | Adams et al, 2011 |
| Automated Primary Care Screening in Pediatric Waiting Rooms | Anand et al, 2012 |
| Advanced Clinical Decision Support for Vaccine Adverse Event Detection and Reporting | Baker et al, 2015 |
| Improved Documentation and Care Planning with an Asthma-specific History and Physical | Beck et al, 2012 |
| Impact of a Computerized Template on Antibiotic Prescribing for Acute Respiratory Infections in Children and Adolescents | Bourgeois et al, 2010 |
| Improving Immunization Delivery Using an Electronic Health Record: The ImmProve Project | Bundy et al, 2013 |
| Use of a Computerized Decision Aid for ADHD Diagnosis: A Randomized Controlled Trial | Carroll et al, 2013 |
| Use of a Computerized Decision Aid for Developmental Surveillance and Screening: A Randomized Clinical Trial | Carroll et al, 2014 |
| Impact of Real-time Electronic Alerting of Acute Kidney Injury on Therapeutic Intervention and Progression of RIFLE Class | Colpaert et al, 2012 |
| Recognizing Hypoglycemia in Children Through Automated Adverse-event Detection | Dickerman et al, 2011 |
| Impact of Clinical Alerts Within an Electronic Health Record on Routine Childhood Immunization in an Urban Pediatric Population | Fiks et al, 2007 |
| Impact of Electronic Health Record-based Alerts on Influenza Vaccination for Children with Asthma | Fiks et al, 2009 |
| Improving Adherence to Otitis Media Guidelines with Clinical Decision Support and Physician Feedback | Forrest et al, 2013 |
| Electronic Health Record Identification of Nephrotoxin Exposure and Associated Acute Kidney Injury | Goldstein et al, 2013 |
| Developing Clinical Decision Support Within a Commercial Electronic Health Record System to Improve Antimicrobial Prescribing in the Neonatal ICU | Hum et al, 2014 |
| Development and Performance of Electronic Acute Kidney Injury Triggers to Identify Pediatric Patients at Risk for Nephrotoxic Medication-associated Harm | Kirkendall et al, 2014 |
| A Quality Improvement Project to Improve Compliance with the Joint Commission Children’s Asthma Care-3 Measure | Kuhlmann et al, 2013 |
| Development of a Web-based Decision Support Tool to Increase Use of Neonatal Hyperbilirubinemia Guidelines | Longhurst et al, 2009 |
| Using Electronic Health Record Alerts to Provide Public Health Situational Awareness to Clinicians | Lurio et al, 2010 |
| Impact of a Clinical Decision Support System on Antibiotic Prescribing for Acute Respiratory Infections in Primary Care: Quasi-Experimental Trial | Mainous et al, 2013 |
| Optimizing Care of Adults with Congenital Heart Disease in a Pediatric Cardiovascular ICU Using Electronic Clinical Decision Support | May et al, 2014 |
| Embedding Time-limited Laboratory Orders Within Computerized Provider Order Entry Reduces Laboratory Utilization | Pageler et al, 2013 |
| Use of Electronic Medical Record-enhanced Checklist and Electronic Dashboard to Decrease CLABSIs | Pageler et al, 2014 |
| Impact of Electronic Medical Record Integration of a Handoff Tool on Sign-out in a Newborn Intensive Care Unit | Palma et al, 2011 |
| Integrating the Home Management Plan of Care for Children with Asthma into an Electronic Medical Record | Patel et al, 2012 |
| Integration of Clinical Decision Support with On-line Encounter Documentation for Well Child Care at the Point of Care | Porcelli & Lobach, 1999 |
| Childhood Obesity: Can Electronic Medical Records Customized with Clinical Practice Guidelines Improve Screening and Diagnosis? | Savinon et al, 2012 |
| Putting Guidelines into Practice: Improving Documentation of Pediatric Asthma Management Using a Decision-making Tool | Shapiro et al, 2011 |
| Optimization of Drug-Drug Interaction Alert Rules in a Pediatric Hospital’s Electronic Health Record System Using a Visual Analytics Dashboard | Simpao et al, 2015 |
| Developing and Evaluating a Machine Learning Based Algorithm to Predict the Need of Pediatric Intensive Care Unit Transfer for Newly Hospitalized Children | Zhai et al, 2014 |
| Improving Home Management Plan of Care Compliance Rates Through an Electronic Asthma Action Plan | Zipkin et al, 2013 |
| Prospective Trials and Studies | |
| Electronic Health Record-based Decision Support to Improve Asthma Care: A Cluster-randomized Trial | Bell et al, 2010 |
| A Shared E-decision Support Portal for Pediatric Asthma | Fiks et al, 2014 |
| Adoption of Electronic Medical Record-based Decision Support for Otitis Media in Children | Fiks et al, 2015 |
| Comparative Effectiveness of Childhood Obesity Interventions in Pediatric Primary Care: A Cluster-randomized Clinical Trial | Taveras et al, 2015 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
