Article Critique by Bolu Oluwalade

Article Critique by Bolu Oluwalade.
Topic: An analysis of electronic health record-related patient safety concerns
Authors: Derek W Meeks, Michael W Smith, Lesley Taylor, Dean F Sittig, Jean M Scott, Hardees Singh

Meeks, Derek W., et al. “An analysis of electronic health record-related patient safety concerns.” Journal of the American Medical Informatics Association 21.6 (2014): 1053-1059.
This study is a retrospective analysis of completed investigation reports about EHR related safety concerns. The purpose of the study was to analyze the electronic health record for patient safety concerns. To achieve this, the authors analyzed the completed investigation reports of the Informatics Patient Safety Office of the Veteran Health Administration (VA) which maintains a nonpunitive, voluntary reporting system to collect and investigate EHR-related safety concerns. The VA operates the largest integrated health system in the USA with over 1700 sites of care. With the advent and wide adoption of the EHR, safety issues have been closely monitored due to the delicate nature of the health care system, a safety issue could potentially lead to the loss of lives hence research in this area is encouraged within the healthcare space. For this study, the authors extracted 100 consecutive, unique and closed investigations between August 2009 and May 2013 from 344 reported incidents. The sample size was chosen based on the review of previous explanatory studies in patient safety which has been thoroughly informative. The narrative data was then analyzed by the authors using a framework analysis method which allows emerging trends to be incorporated into the previously established framework. The result obtained from the analysis shows EHR-related safety concerns involving both unsafe technology and unsafe use of technology. The authors concluded that institutions with long-standing as well as recently implemented EHR systems should build a monitoring system that examines these safety concerns and learn from them.

Since the displacement of the paper-based health record system by the EHR, there has been notable improvements in healthcare delivery, it’s is more coordinated, effective and productive. But as we continue to develop and use the EHR, we need to constantly learn about these human-computer interactions. To avoid errors and safety concerns we must develop the right technology, use it safely and develop a monitoring and control system that can help us evaluate how we can make the best use of it. The background of the study discusses the benefits of health IT most especially the Electronic health records. It also addresses the safety concerns that comes with the wide adoption and use of the EHR. The authors account for the significance of their study on the fact that detecting and preventing EHR related safety is challenging because of its multifaceted nature hence the need for new sociotechnical approaches that account for these various elements. They highlighted the reasoning behind the use of their previously developed sociotechnical model that proposes eight independent dimensions that are essential in the understanding of EHR-related safety concerns.

Design and setting
The authors performed a retrospective analysis of completed investigation reports about EHR related safety conducted by the informatics Patients Office of the Veteran Health Administration(VA) which operates the largest integrated healthcare system in the US with over 1700 sites of care. The VA developed the informatics Patients System office to establish a nonpunitive, voluntary reporting of EHR related safety concerns within its network. The IPS, though is limited to only IT-related reports. The reports submitted to the IPS are voluntary and not mandatory so there is a chance that not all safety issues are submitted. The most common ways of reporting involved the EHR users reporting any safety concerns to the local IT team, who in turn analyze the concerns and relates if needed to the national help desk. The authors did not mention if there are any reporting guidelines followed by the local IT team to determine what goes to the national help desk . At the national level the IPS analyst investigates the reports, it takes about 30 days for the investigation to be completed. The IPS analyst develops a score according to potential severity, frequency, and detectability. The safety issues are addressed based on these scores. After the investigations, IPS indicated the investigation as closed. These closed reports were the input data used by the authors.

Data collection
The authors searched the IPS database for the closed investigation reports that contain full analysis and provided meaningful information. The authors, by excluding input errors such as erroneous editing, and solely focusing on only health IT concerns reduced the scope of the study. An appropriate representation of the research scope in the topic would have been “An analysis of health IT-related safety issues in the EHR”.
The data was collected within a single health system thus the safety concerns identified could be limited to that system. The use of different systems would have led to an increased range and a more representative data.
The reporting of the various safety issues by the EHR system users was not mandatory hence it is uncertain how often the user’s reports errors with the system, from my personal experience with the EHR, most often if errors did not directly lead to patient harm, it may not be reported. The authors did not state if there is standard format of reporting errors that were identified by the users. A standard error report format would facilitate the collection of accurate and adequate information hence helping with the investigations of such errors.
Data Analysis
Narrative data from the closed IPS investigations were analyzed using the framework analysis technique which consists of five stages: familiarization, thematic analysis, indexing, charting and mapping and interpretation. I agree that the data analysis technique used by the author is appropriate for narrative data. The familiarization was done by two of the authors who reviewed and studied the closed IPS reports. The thematic analysis was done using the eight-dimension sociotechnical model with a coding technique developed for each concern in the model. In addition, the authors categorized the concerns into three different phases: phase 1 concerns are related to inherent unsafe technology or technology failures, phase 2 are concerns related to the unsafe and inappropriate use of technology, phase 3 are concerns related to lack of the use of technology to monitor potential safety concerns before harm occurs. Hence the coding technique allowed the authors to classify safety concerns into multiple dimensions from the sociotechnical model but only in one EHR safety phases the coding authors independently indexed safety concern after reviewing closed IPS investigations.
The authors generated emergent safety concerns through collaborative and iterative analysis of the whole set of coding results and charted to allow groupings of reports that represented a common theme. The emergent and recurring safety concerns were mapped and interpreted according to their sociotechnical origin and EHR safety phase.
According to the authors, “Most often, nontechnical dimensions such as workflow, policies, and personnel interacted in a complex fashion with technical dimensions such as software/hardware, content, and user interface to produce safety concerns.” Based on this, I agree that the choice of the sociotechnical approach was appropriate.

The authors extracted 100 consecutive, unique, closed IPS investigation reports from 344 reported between August 2009 and May 2013 reported from 55 different VA facilities. 48 had a low score, 38 intermediate scores, 14 scored high. About three quarter are classed as phase 1(related to unsafe technology use), one quarter was in phase 2(unsafe EHR use), with just one safety concern reported for phase 3(failure to use the EHR to monitor patient safety). The author’s classifications were appropriate as it covers the major safety Health IT concerns reported. The authors reported the occurrence of more than one sociotechnical dimension: forty had two, twenty-three incidents had three and seven involved four dimensions. The authors during chatting, mapping, and interpretations of social and technical components of the EHR use identified several distinct safety concerns and classified them into four types which are : unmet display needs in the EHR(this represented when the human-EHR interaction did not adequately support the tasks of the end users hence potentially leading to uncertainty of the users or damage to the patient), safety concerns with software modifications or upgrades, concerns related to data transmissions at system-system interfaces and concerns of hidden dependencies in distributed systems which is basically when one component with the EHR unknowingly affects the state of another component. Although all the four types were properly defined and classified in the table, the lack of a diagrammatic visualization of their findings makes it more difficult to understand. It would have been more presentable if the relationship between sociotechnical safety phases and safety concerns types were presented in a diagram such as charts, graphs or plots.
The authors highlighted the importance of their findings which was that it could be useful in developing ways to monitor and improve safety. The authors identified the advantage of their novel approach to this study which is the sociotechnical study of an information-rich data source over previous studies which have used isolated events without the benefit of an independent factor and informatics investigation to identify safety concerns in the EHR. This has proved to be an adequate technique and has been adopted by some subsequent studies
The findings of the authors demonstrated the significance of establishing a monitoring system for the detection of errors and safety issues in EHR after it’s installation. A reporting and investigation infrastructure like that of the VA is also needed for larger institutions. The authors also concluded that proactive risk assessment will go a long way to maintaining a safer use of EHR. The authors also suggested the involvement of people, organizations, systems, technology, and policies that influence how the EHR is used, should enable for a better EHR system. These findings shed more light on the issues of safety in the EHR and have served as a foundation for further studies.

The study identified and classified several EHR related patients’ safety and its findings has served as a source of guidance for further investigations. There are some areas that I believe the study could have been improved on. Considering the scope of the study, the topic could be better defined. The authors did not state if there were any external factors that could lead to safety concerns for patients e.g. malware or hackers who could potentially gain access to the EHR and cause damages. The data collected by the IPS which is reported by users can be affected by the reporter’s bias or knowledge, if there was a mandatory error report coupled with the standardized format of reporting the EHR safety concerns, the validity of the research would have been greatly enhanced. Although the right analysis method was carried used, the sample collected was not shown to be representative of the large population of US in numbers and diversity. The authors did not use diagrams which could have better depicted their findings and relationships between the defined sociotechnical phases and safety concerns types. The authors did not identify which healthcare units generated the most EHR-related safety concern i.e. if the emergency units generated more errors than the radiology units using the Radiography information systems (RIS) or laboratory units using the laboratory information systems( LIS). They also did not identify the types of EHR-safety concerns generated by the various sites of care e.g. hospitals, clinics, community living centers etc. This could have been an important information.
Finally, the authors did a good job in analyzing the safety issues within the EHR used by the veteran health administration, but this cannot be said to be a representative of most of the safety issues with the various EHR system deployed within various healthcare networks in the United States. Despite some limitations, this study serves as a significant step in encouraging analysis into various EHR systems for patient safety concerns.