Over the last 50 years, the electronic health record (EHR) has emerged as a critical tool in the delivery of safe, efficient, quality healthcare. EHR systems have evolved from basic single-office databases into sophisticated applications capable of managing clinical documentation, laboratory results, images, and other patient data across care settings as well as providing decision support to promote safe patient care, reduce errors, and support adherence to practice guidelines. Data captured within the EHR are used to support nearly every aspect of patient care, including related billing and auditing activities. Increasingly, these data are also being used for quality improvement and research.

Hospitalists and the patients whom we treat stand to benefit greatly from well-implemented EHRs that provide tools to review growth data and immunization histories, to identify when vital signs and laboratory values exceed normal parameters for age, and to deliver age-, weight-, and condition-appropriate decision support for medication dosing and management. However, most commercially available EHR systems were designed with adult patients in mind. Configuring these systems to care for children often requires additional customization, which can translate to the need for hospitalists to invest time working with EHR implementation teams to ensure that safe, efficient, quality pediatric care can be delivered. This is especially important for hospitalists who work in pediatric units within larger adult-centered hospitals. In this chapter, we provide a brief background of EHR systems, discuss the role of clinical decision support (CDS) tools in delivering safe, efficient, quality pediatric hospital care, and review important patient safety considerations for hospitalists who may be asked to participate in the design, implementation, or optimization of an EHR.

HISTORY OF HOSPITAL MEDICAL RECORDS

The practice of keeping medical records is one of the cornerstones of medicine. Hippocrates, the ancient Greek physician and father of Western medicine, famously kept and advocated for the use of medical records as early as the fifth century BC.¹ These earliest medical records were chronological accounts of individual patient cases and served as important tools in the initial understanding of the natural history of diseases, patient outcomes, and sharing of knowledge among practitioners. Such physician-based records dominated medical documentation for centuries and relied on the diligence of the conscientious physician to accurately document, maintain and preserve this information.

In the mid-nineteenth century, the emergence of record keeping in hospitals represented an early attempt to organize patient information primarily for tracking cases and billing.² Some saw the additional value of the aggregate information from the hospital population for improving patient care. In 1863, social reformer, nurse, and statistician Florence Nightingale described the challenges presented by the limitations of medical record keeping at that time:

The call for structuring and standardizing the medical record to improve its usefulness for healthcare and research grew stronger in the twentieth century, at the same time the computer was emerging as a tool for business and research. In the 1960s, physician Lawrence Weed proposed transitioning to a problem-oriented medical record to improve the ability of clinicians to identify, manage, and treat a patient’s problems in the context of a shared record, as well as to measure and advance the quality of care delivered. He advocated for the use of the computer as a tool to provide better access to the medical record and help guide care delivery.^4,5 Weed’s concepts helped guide the design of early EHRs, and some of these concepts are now standard components of medical documentation.

STRUCTURE AND COMPONENTS OF MODERN EHRs

Modern EHRs continue to evolve from their origins as basic databases into sophisticated applications capable of managing clinical documentation, laboratory results, imaging, and other patient data across care settings. Although many use the terms EHR and electronic medical record (EMR) interchangeably, there is a distinction: EMR classically refers to the electronic version of the paper chart contained in a single office, while EHR refers to the patient-based record of care across multiple care settings and providers.⁶ In 2003, the Institute of Medicine (IOM) developed the following definition of the EHR:

To further illustrate this definition, the IOM broadly outlined EHR core functions, shown in Table 6-1.

A hospital’s EHR provides these functions by integrating numerous components.⁸ The Admission, Discharge and Transfer (ADT) component manages patient location and identifying information. The Computerized Physician Order Entry (CPOE) component enables the entry of orders that can then be communicated to different staff members or to specific hospital departments like the pharmacy, radiology, and laboratory. The Laboratory Information System (LIS) component handles laboratory result data transmission, and a Picture Archive and Communication System (PACS) component manages radiographic images and reports. Additional components manage non-clinical functions, including billing and accounting services.

PEDIATRIC-SPECIFIC EHR FUNCTIONS

The care of children presents many unique challenges for EHR systems. The American Academy of Pediatrics’ (AAP) Council on Clinical Information Technology identified many key EHR features for pediatricians practicing in ambulatory and inpatient settings.⁹ These features include immunization support, growth tracking, medication dosing, and age-based data norms (Table 6-2). Among the most often cited concerns surrounding pediatric EHR adoption is the lack of robust functionality within commercial systems for providing pediatric care.^10-13 Most pediatric-specific features are not automatically built-in when a hospital purchases an EHR, and those that are included often require detailed configuration by teams of clinicians, pharmacists, and the EHR vendor. During this process, special attention must be paid to the complexities of caring for patients whose care needs, normal reference ranges for vital sign and laboratory data, and recommended best screening and treatment practices can vary greatly and change rapidly with age, body weight, physiology, and developmental level.

To further address the need for standards in the development of EHRs that could help meet the needs of children, the Agency for Healthcare Research and Quality (AHRQ) recently developed the Children’s EHR Format.¹⁴ The goal of the format was to provide the many key stakeholders, including EHR developers, EHR purchasers, and the end user care providers, with an understanding of the minimum requirements for data standards and EHR features to optimize pediatric healthcare. These lists of features by the AAP and the AHRQ can serve as a useful starting point for hospitalists evaluating potential new EHR systems or optimizing the features of an existing system.

CURRENT STATE OF EHR ADOPTION IN CHILDREN’S HOSPITALS

Although the first EHR systems were implemented in the 1970s, a recent survey of children’s hospitals revealed that only 17.9% had a basic EHR system, and only 2.8% met criteria for having a comprehensive EHR.¹² While over 95% of hospitals had radiology images and lab results available electronically, fewer had drug-allergy alerts (62%), age-based dosing support (44%), CPOE (34%), problem lists (24%), and physician notes (13%). Financial costs were identified as the largest barrier to EHR adoption or comprehensive feature implementation. Recent efforts by the United States government to provide financial incentives for EHR use may stimulate advancement in these low adoption rates.

FEDERAL INCENTIVES FOR EHR ADOPTION

In an effort to spur adoption of health information technology, the federal government allocated 155 billion dollars in the American Recovery and Reinvestment Act of 2009 (ARRA) to fund Title XIII, the Health Information Technology for Economic and Clinical Health (HITECH) Act. The HITECH Act allocates nearly 26 billion dollars to incentivize the adoption and “Meaningful Use” of certified EHR systems, as strictly defined by the Office of the National Coordinator for Healthcare Information Technology (ONCHIT). These incentives are in the form of Medicare and Medicaid payments to eligible professionals and hospitals.^15,16

The three stages of Meaningful Use are designed to help organizations guide their incremental implementations of EHR systems.^17,18 The goal of Stage 1 is to implement an EHR capable of data capture and sharing, followed by Stage 2’s requirement to demonstrate more advanced clinical processes and finally Stage 3’s requirement to show improved patient outcomes¹⁹ (Table 6-3). Hospitals must attest that their EHR meets certain quality measures for certification in order to advance to the next stage. To aid providers in undertaking EHR implementations, the HITECH act also describes funding for the creation of Regional Extension Centers (RECs) to provide information technology support, as well as other workforce training and research programs in information technology.¹⁷

Among the most anticipated benefits of the modern EHR is its potential not only to serve as a repository of clinical data and billing information, but also to assist in diagnostic and therapeutic decision-making. By translating complex medical knowledge into evidence-based rules to be applied in specific patient care situations, these systems can help physicians provide higher quality care. Tools that can help physicians prescribe the right doses of medications, avoid dangerous drug-drug interactions, and guide evidence-based treatment decisions are just a few examples of what is collectively known as clinical decision support (CDS).

DEFINITION AND FORMS

CDS is “knowledge- and person-specific information, intelligently filtered or presented at appropriate times, to enhance health and health care.”²¹ Although CDS existed long before the EHR in the form of paper-based templates of common orders for specific conditions, clinical algorithms, drug references, and treatment guides, these all required a provider to actively seek out information. Ideally, an EHR’s CDS system is able to retrieve pertinent clinical information and offer it to the clinician in a clear and concise manner in order to maximize the efficiency and effectiveness of the decision-making process. Forms of CDS include reminders of routine screenings or vaccinations, alerts for critical situations, CPOE order sets, clinical guidelines, and user-friendly displays of general or patient-specific diagnostic and therapeutic information.²²

Any CDS intervention should be viewed in the context of the “CDS Five Rights” to ensure delivery of the right information to the right person, in the right format, through the right channel, at the right time²³ (Table 6-4). These rights serve not only to maximize the effectiveness of CDS, but also to reduce the likelihood of unintended consequences of CDS that may be detrimental to patient care. A review of randomized controlled trials suggested that CDS systems improved clinical practice in 68% of studies, and the four independent predictors of improvement were the use of a computer-based system, the automatic provision of CDS, provision of recommendations rather than assessments, and the delivery of these interventions at the time and place of decision-making.²⁴ Common forms of CDS are reviewed and summarized below, with specific reference to the benefits and challenges of medication alerting, order sets, and clinical practice guidelines in a hospital setting.

EXAMPLES

Medication Alerts

With estimates of 44,000 to 98,000 deaths attributable to medical errors annually, a significant proportion of which are related to medications, a clear potential benefit of CDS is to reduce the number of preventable Adverse Drug Events (ADEs).²⁵ This potential benefit has been studied in both adult and pediatric settings, and many analyses have demonstrated statistically significant reductions in prescription errors and some reductions in ADEs.^26-30 CDS interventions for medication order completion can range from fairly simple allergy and drug-drug interaction checking to more advanced support such as formulary-based dosing shortcuts, maximum dose alerts, indication-based dosing, and integration of pertinent patient data such as lab results. Ideally a system has tools to prevent errors of commission, when the action taken is incorrect like medication overdose, as well as errors of omission, when a needed therapy is not ordered, such as serum medication levels in a patient with poor renal function.³¹

In pediatrics, age- and weight-based dosing are the primary methods for prescribing medications. Variations in these patient characteristics can be gradual and predictable as in the yearly well-child visits at the pediatrician’s office, or volatile as in the daily fluctuations seen in a neonatal intensive care unit. Performing dose calculations is perhaps the most fundamental of pediatric CDS functions. In addition to performing calculations, the EHR should also support the physician in ensuring that the weight used is the most accurate and up-to-date as possible. Weight accuracy can be supported by comparing the recorded weight to standard norms (Centers for Disease Control [CDC] growth curves) and by alerting the clinician of significant deviation, as well as alerting for implausible changes in weight over short periods of time.

In addition to supporting calculation of an appropriate medication dose, a CDS system should alert a prescriber when an ordered dose is too low or too high for the patient. This should include both weight-based dose checking and absolute maximum dose checking. More advanced systems would also include support for appropriate dosing by indication (e.g. different antibiotic doses for pneumonia vs meningitis) or by co-morbidity (e.g. recommending a modified dose for patients with renal insufficiency) by either prompting the user to specify these conditions or querying the patient’s problem list. In addition, the format of alerts can vary with escalating levels of interruption and override requirements based on the severity of the alert.

Order Sets

Order sets in CPOE systems group associated orders for specific clinical situations, conditions, or workflows for rapid ordering and can help guide the delivery of safe, efficient, quality care. Examples include orders grouped for specific presenting symptoms, such as fever in young infants, for conditions such as asthma, or for workflows such as general pediatric admissions. Order sets can also be used to suggest corollary orders (e.g. serum antibiotic levels to be ordered with aminoglycoside antibiotics, or urine pregnancy tests to be ordered prior to radiology studies for female patients). Order sets constitute a form of CDS in that they present the most appropriate therapeutic interventions and diagnostic testing for a given situation, reflecting the accepted standard of care of an institution.²² The benefits of using order sets include improving the consistency of care delivered, as well as helping to decrease the cognitive workload on providers.^32,33 In addition, by guiding clinicians to the correct initial decisions, order sets can reduce the number of workflow interrupting alerts faced later in the ordering process.

The ability to create order sets is a feature present in most EHRs, and there are many guidelines and reviews of best practices for their design and effective implementation.^34-37 Although commercial EHRs may include pre-packaged “pediatric” order sets, it is important to check that they are clinically appropriate and fit into the workflow of the providers who will use them. A well organized, evidence-based order set may be of little use to providers if it does not fit into their actual workflow, is not readily accessible at the right time in their workflow, or if they are not aware of its existence.^22,38

Clinical Practice Guidelines

A clinical practice guideline (CPG) represents an evidence-based, systematically developed set of recommendations for the management of a specific condition.³⁹ The use of such guidelines is intended to improve healthcare quality and decrease variation in care delivery. A review of barriers to physician adherence to CPGs highlighted the lack of provider awareness of guidelines, as well as limited time to access them and absence of reminders to use them.⁴⁰ Such barriers can be overcome by integrating CPGs into the EHR: translating the CPGs into rule-based alerts to identify clinical situations where guidelines are applicable, and using order sets to organize recommended tests and treatments. Such guideline-related CDS tools have been demonstrated in modern EHR systems and were shown to increase guideline adherence and positively influence clinical practice.^41-43

Multiple challenges to the successful implementation of CPGs in the her exist. Many CPGs are long, complex documents and not suitable for onscreen display in their entirety. Efforts at simplifying guideline information while still providing access to the details of the full guideline is important in improving physician adoption.³⁴ CPGs can require multiple pieces of patient data to aid in decision-making. The more automated the data collection process, and the less the user must manually enter, the more likely they are to use the system. Guideline maintenance is also very important, as the standard of care changes with time and so they should be reviewed regularly. As the EHR is a centrally managed system, rapid deployment of updates is possible and does not require the replacement of prior paper guidelines, thus avoiding the costs and issues of version control. This approach helps reduce system variation and can be a powerful tool in helping avoid errors and improving quality of care.⁴⁴

Although the benefits of EHR adoption include mechanisms to improve safety and quality in the healthcare system, the EHR itself is not without its own safety risks. As the technology behind the EHR grows more sophisticated and intertwined with the operations of a healthcare organization, it is increasingly difficult to anticipate potential issues between the EHR and providers.⁴⁵ An EHR will do exactly what it is programmed to do, but flaws in system design or integration into provider workflow can carry unintended consequences and adversely affect patients.⁴⁶ In the rush to implement EHRs to meet Meaningful Use criteria, healthcare organizations without EHR experience are at higher risk for such consequences. Sittig and Singh recently called attention to the safety challenges associated with EHR technology, including the adverse impacts of downtime and the unintended consequences of failing to use EHRs appropriately.⁴⁷ These are outlined briefly below.

DOWNTIME

As an EHR implementation steadily progresses towards the ideal of a fully electronic, paperless system, a healthcare organization’s operations will become more and more reliant on the EHR being up and running without interruptions due to unplanned crashes or routine maintenance. Downtimes expose weaknesses in the organization’s network and EHR technology infrastructure, while forcing clinicians to deliver medical care without the safety net of clinical decision support that they may rely upon.^48,49 The need to resort to paper-based documentation and ordering processes can have a major impact on efficiency and patient safety, particularly as each successive generation of care providers will be less and less familiar with practicing without the aid of a computer. The effect of lengthy downtimes on an entire healthcare system can be profound, slowing care processes to the point that it is difficult to safely assume care of new patients and leading to considerable losses in revenue.

Even the most secure and highly reliable system is vulnerable to an unplanned downtime, and every organization should be ready for the possibility. Steps to prepare for downtimes include establishing a downtime policy and contingency plans for extended downtime periods.^50,51 Support materials like paper forms for documentation and ordering, drug and laboratory reference books, and backup communication systems should be in place to ensure care can continue without the EHR. Planning must include not only for the downtime itself but also the recovery process to return to the pre-downtime state as well as activities like the re-entry of orders and clinical data into the EHR to ensure completeness of the medical record. Such operational and recovery plans are only as effective as provider familiarity with the plans, so providers should be trained on downtime procedures, even practicing them as one might practice a fire drill or other disaster preparedness exercise.

UNINTENDED CONSEQUENCES

How an organization chooses to implement and use its EHR system can have major impacts on patient safety and workplace efficiency. Underestimating the need for clinical process requirements gathering, redesign, and validation during an EHR implementation can have direct adverse effects on patients. As an often-cited example, Han et al. demonstrated that implementation of a new CPOE system was associated with an unexpected increase in mortality.⁵² The authors described difficulties in placing multiple orders quickly and efficiently, the inability to pre-order medications for critically ill patients en route to the hospital due to registration workflows, and delays in obtaining medications in emergent situations as contributing factors. However, their report was met with stiff criticism, and can serve as a case study on the effects of failing to appropriately plan for EHR implementation.^36,53 As an example, carefully designed ICU-specific order sets could have reduced the time required to order urgent medications. Consideration of a “pre-registration” process could have allowed the organization to enter orders on patients who were en route to the intensive care unit. Furthermore, the authors also noted that elimination of a satellite medication dispenser located in the ICU led to delays in medication availability, an issue that could have been detected through EHR workflow simulations prior to go-live. Learning from this experience, subsequent implementations at other hospitals have described successful CPOE system implementation without adversely impacting mortality rates.^53,54

Even the best intentions to leverage and maximize the capabilities of an EHR’s CDS systems can introduce and facilitate new errors. CPOE systems can easily check the list of active medications for possible drug-drug, drug-allergy, drug-food, and drug-condition interactions. Given a streamlined and clinically validated set of possible interactions, these systems can help prevent administration of potentially dangerous medications. However, when used indiscriminately, these same tools can lead to excessive alerting and overwhelm providers.^46,49 The provision of too many alerts that are frequently overridden by prescribers can lead to “alert fatigue” and lead prescribers to unintentionally ignore important alerts.⁵⁵ At the heart of this problem is that standards and best practices for alert content are lacking. Although work is underway to identify priority drug-drug interactions for adult and pediatric populations, it will be some time before these standards are easily implemented into vendor EHR systems.^56,57

A related example of an alerting tool’s unintended consequences is the use of the “hard stop” alert that cannot be overridden. Such an alert may be meant to prevent a rare drug-drug interaction, but this could be of less risk than the risk of not giving the two medications in some clinical situations. Furthermore, in a time-critical situation, a “hard stop” may be inappropriate and lead to potentially dangerous delays in care.⁵⁸ These and many other examples of unintended consequences of EHR implementation and usage are, unfortunately, common. In response, the AHRQ has created a report and website geared to both future and current EHR users to help them identify and address these issues.^59,60

The EHR is a powerful tool for improving clinical care and augmenting the abilities of hospitalists to deliver that care safely and efficiently. As the use of EHR systems continues to increase, these systems will become the standard form of medical record keeping and care management. Learning how to maximize their potential will be critical for all healthcare providers. Given that the volume of patient health data is expected to grow more rapidly with the emergence of genomic and personalized medicine, the EHR will take on even greater importance in helping healthcare providers organize and interpret this information in the coming years.