Vincent X. Liu, MD, MS; Nimah Haq, MPH; Ignatius C. Chan, MD; and Brian Hoberman, MD, MBA
Electronic health record (EHR) implementation within US acute care hospitals increased rapidly between 2010 and 2015 as a result of the Health Information Technology for Economic and Clinical Health Act of 2009.1-3
Among hospitals with and without an EHR system in 2009, some of the major barriers to EHR adoption were perceived to include the cost of purchase, uncertain return on investment, clinician resistance, and inadequate information technology staff.4,5
Many of these barriers have been persistently cited as concerns during the subsequent period of rapid EHR uptake.6-18
Ongoing maintenance has also been cited as a key barrier to EHR implementation and use. However, few studies have described the subsequent improvements needed to maintain and optimize EHR functionality over time.4,5
Given the substantial costs resulting from initial implementation, far less attention has been devoted to the maintenance needs that follow for optimal EHR performance. In this study, our goal was to describe the scale and scope of changes made to an inpatient EHR system after initial implementation within the 21 hospitals of Kaiser Permanente Northern California (KPNC).
This study was deemed exempt by the Kaiser Permanente Institutional Review Board. KPNC is an integrated healthcare delivery system serving 4.1 million members at 21 hospitals. Implementation of an inpatient EHR system (Epic; Verona, Wisconsin), known internally as KP HealthConnect (KPHC), was completed in 2010; outpatient KPHC completed implementation in 2008.19
We evaluated significant changes made to the inpatient EHR system based on documentation within monthly “KPHC communication” reports between 2010 and 2015. These reports were used to inform end users about upcoming KPHC updates (ie, changes to existing functionality) and/or upgrades (ie, new functionality). Multiple unique changes affecting the same EHR functionality were included within a single change document, which described the overall changes to that functionality.
The changes arose from diverse sources, including clinician requests; feedback from clinical technology leads; safety, risk, or reporting needs; and/or forthcoming system upgrades. Although several governance committees exist to oversee EHR changes, the modifications described here were shepherded through the change management process by a core regional team of physician and clinician informaticists with significant experience in EHR clinical functionality; working alongside this core team was informatics staff experienced in building EHR tools. The changes documented within these reports represented only a subset of all requested changes, as not all requests were ultimately fulfilled. Thus, the reports included only those changes that were implemented as prioritized by clinical value, safety, regulations, resource availability, and end-user satisfaction. Documented changes also varied in scope and scale, with some being implemented with relative ease and others requiring months of preparation.
After identifying all changes grouped within the documents, we further categorized them within broad functional domains based on their type, including orders (single, grouped, and templated orders), alerts and customization (user-customized tools and decision support), surgical and emergency department (ED) (perioperative- and ED-specific tools), data review (clinician-facing tools for review of patient charts, results, and imaging), reports and health information management (HIM) (unit- and group-based reporting tools), and other (patient tools and other change types). For each change, we also identified and grouped the type of user affected, as well as the clinical or functional area affected, based on report documentation.
Data are reported as median (interquartile range [IQR]) and number (percent). Analyses were conducted using STATA 14.1/SE (StataCorp; College Station, Texas).
Between 2010 and 2015, 5551 unique changes were made to the inpatient EHR (Figure 1
), with a median of 72 (IQR, 35-112) changes per month. Most unique changes (n = 3191 [57.5%]) were updates to existing functionality, with 95.7% affecting all 21 hospitals. Individual changes were aggregated within 2190 update communication documents.
Upgrades related to EHR orders contributed to the largest proportion of all significant changes (44.7% of documents) (Figure 2
). In total, changes to templated order sets comprised 29.9% of all documents. Other EHR functional domains that accounted for a significant proportion of all changes included clinical data review (15.7%), surgical and ED-specific tools (13.5%), alerts and customization (11.4%), grouped reports and HIM (8.3%), and patient tools and other (6.3%). Overall, changes affected 135 EHR functions.
In total, 151 specific types of users were affected by changes (eg, bed controller, cardiologist, certified nurse midwife, support site specialist), with an impact on all KPHC users 10.2% of the time. Targeted changes most frequently affected nurses (30.6%), physicians (26.6%), and other clinical staff (22.7%), such as pharmacists, therapists, and dietitians. The specific clinical areas most commonly affected by changes included surgical specialties (7.9%), ED (7.1%), mother–baby (6.9%), and pharmacy (6.4%) (Table
Over a 6-year period, the changes required to maintain and improve an inpatient EHR system were substantial and diverse, with a pervasive impact. On average, 2.5 significant EHR changes occurred each day, together affecting more than 130 specific tools and 150 unique user roles across the 21 hospitals. Key areas that were frequently targeted by updates included specific EHR tools (order sets), clinical domains (surgical and ED), and end users (nurses, physicians, and pharmacists).
Widespread implementation of inpatient EHR systems has occurred rapidly over the past decade; however, few studies have comprehensively detailed the maintenance required to optimize their use.4,5
Most studies evaluating EHR implementation have focused on quantifying the costs and barriers related to initial implementation, the outcomes associated with EHR uptake, and the impact of EHR use on clinicians and patients.6-18,20-25
Nevertheless, emerging evidence suggests that the benefits of EHR use accrue gradually over time and are likely attributable to the ongoing addition of new functionality attained via continual updates and upgrades.21,26
Thus, although much attention is focused on the initial “go live” of the system, the true benefits of EHR adoption may only emerge with persistent attention to enhancing the EHR-based workflows and tools that drive improvements in care. In particular, the customization and usability of EHR functions to meet end-user needs have been identified as key measures that portend likely EHR benefit and can also mitigate potential harm arising from usability or workflow challenges.27-30
The primary limitation of this study is that our findings were based on a single healthcare system and a specific EHR product, which may limit the generalizability of our findings. Our study focused on the clinical aspects of inpatient EHR maintenance recorded within monthly change communication reports. However, our findings almost certainly represent a significant underestimate of the true scale and scope of ongoing EHR changes across our system. Numerous daily changes are made to EHR functions that do not rise to the significance level that would trigger their inclusion within communication reports. Our findings also do not account for simultaneous outpatient EHR and information technology infrastructure support, which contributes heavily to ongoing maintenance needs. Finally, the resources needed to implement each change could vary significantly, in terms of both time and cost.
EHR maintenance needs were prevalent and diverse, affecting 150 unique user roles and contributing to an average of more than 2.5 significant changes per day. Our findings highlight the need for significant resources, expertise, and collaboration to maximize EHR clinical utility and benefit. They also demonstrate that an EHR system represents a dynamic network of evolving tools that requires ongoing investment well after initial implementation.