myang875
commented
4 years ago Reflection by Mu
I think EHR certainly provides a lot of functionalities over traditional paper-based patient records. If used correctly, they are certainly capable of improving on healthcare quality and reducing medical errors. However, it has pros and cons, and some places still don't adopt EHR, which are influenced by many factors that we could not address. Although the system is definitely far from perfect, it is actually frustrating to think how many medical errors are still happening even when using this system, but I guess what we are trying to learn is to find the deficiency of this system and try to fix them.
pristineliving
amenschik
1. Introduction
Electronic health record (EHR) is a patient medical record stored electronically. It serves much more purpose than information storage, however, and is not a direct transformation of a paper-based patient record. EHR differs from a paper-based patient record in a few ways. 1) EHR allows for a more flexible way to input data and present information. It can take information input in various formats, as the system can easily convert different formats to the same format for storage, and present the information in the format according to users' preference. 2) EHR facilitates more standardized information storage. It can limit the users of their input, forcing them to choose from a drop-down menu, or requiring the input to be one of the controlled vocabularies. This can ensure the data to be stored in a consistent way, which would smooth the way for data processing, should they be needed. 3) EHR offers uncomplicated accessibility. With secure encryption, users can access patient records from not only their office but their home. Multiple users can access patient records at the same time. This is useful for patient handoff, or when different health personnel is looking after the same patient. 4) The system could prevent input errors by providing spell checks or making users choose from the given choices. The digital record may also reduce the error caused by unreadable handwriting. 5) The system could prompt users to take note of certain information if the patient possesses some characteristics or traits. 6) EHR can include data in formats other than text. This includes images or videos, which would not be available to paper-based records. With the advance of medical imaging technologies, these types of formats are now widely used and thus are essential to a patient record nowadays.
Functions of EHR
1) Patient Data Integration Patient data involve different formats. EHR can gather information from texts and medical images. It also collects information from various sources, such as from different clinicians or test results returned from different facilities. EHR include information such as patient demographics, vital signs, medications, diagnoses, orders, progress notes, radiology images, consults, immunization, allergies, discharge summary, lab test results, and medical history. To facilitate integration within EHR, health care providers are encouraged to use code standards for communication, such as Logical Observation Identifiers Names and Codes (LOINC), Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT), Unified Code for Units of Measure (UCUM), RxNorm and RxTerms. Health Level 7 (HL7) is used to send and receive clinical data, and there are interface for translation of code if necessary.
2) Computerized Provider Order Entry (CPOE). CPOE lets providers enter orders directly into an electronic system. Orders can be sent off to pharmacies or labs without intermediate processes. This can reduce errors and improve on safety issues. When a clinician prescribes medication in a paper-based patient record, it will go through the following steps: clinical orders are given by the clinician, the prescription is transcribed by a pharmacist, the pharmacist checks for potential risk of allergies and drug-drug interactions and dispenses the drug, finally, the medication is administered to the patient. Each of these steps has a potentiality for error, and CPOE can minimize the error at the ordering and transcribing stage. When entering the clinical order, the system can trigger an alert for drug-drug interactions and allergies. Since prescriptions are sent off to the receiving party directly, the error taking place during transcribing orders can also be saved.
3) Clinical Decision Support (CDS). CDS provides clinicians with knowledge and information to aid decision-making. CDS can improve safety as well as the quality of healthcare and enhance patient outcomes. A suggestion raised by a CDS system, providing a rationale for such an action, is best presented to a clinician where they can react with a simple mouse click. Examples of suggestions include a reminder of a patient's immunization given their condition and age, a recommendation of treatment and their usage, recommendations of follow-up tests or treatments ensuring a test result, and alerts about drug interactions and allergies upon medication orders.
2. Insights of EHR Characteristics and Functionalities
2.1 Benefits and Detriments
For scientists and informaticians, electronic health records (EHRs) offer real-time access to knowledge. They are a source of ongoing and constant data capture; reports of results and aggregated data; and long-term accumulation of clinical evidence, whether pertaining to costs, procedures, or outcomes. EHRs are guides, supporting and improving the clinical decision-making process. Electronic interfaces connected to existing electronic systems, such as pharmacy systems or laboratory systems, provide near-instant availability of clinical data to providers across the healthcare industry.
The capture of the care experience occurs in real-time, with on-demand generation through digital platforms. These experiences can be collected and applied to increase knowledge for care improvement. Thus, EHRs facilitate improvement in patient-clinician partnerships, promoting engagement and empowerment among patients, and aligning incentives by encouraging continuous improvement. Transparency is improved with the implementation of EHR systems, with systematic monitoring of health care standards, such as safety, quality, cost, and outcomes. All these improvements further instill values of continuous learning. Successful data sharing and aggregation promotes teamwork, collaboration, and adaptability.
Despite these benefits, however, there are many features of EHRs that need to be addressed before the full advantages of an electronic system can be understood and shared.
2.1.1 Data Capture
There are two common methods of data capture in existing EHR systems: electronic interfaces connected to existing data sources, and direct manual data entry by physicians. Interfaces between EHRS and existing data sources require significant effort to implement. While the process can be streamlined if an organization is affiliated with both the EHR and the source systems, most often this is not the case, as patient data can be produced and ordered by a myriad of outside organizations – e.g. hospital-based healthcare systems and pediatric private offices. Generally speaking, special procedures and extra work are required to collect all relevant patient data from all sources.
Alternatively, manual data entry can come in three forms: native free-text, coding, or a combination of the two. Coding makes data understandable to a computer, enabling commonly-accessed features such as retrieval, research, system improvement, and data management. Computer interpretation of native free-text requires an integration of native language processing (NLP) systems (see later chapters), which can be used to automatically encode narrative text. They are, however, typically costly and time-consuming. Auto-complete coding schemas can help simplify data input and make the process faster and more efficient.
Physician-entered data currently poses the most difficult challenge to EHR system developers. The documentation burden on physicians has steadily risen over time as patient problems become more acute, care teams grow larger, physicians order more tests, and billing bodies demand more documentation. The time cost of physician input is high, and while it is the belief of some that the person who collects data is responsible for entering it – thus, physicians should invest time in transferring all notes, orders, codes, and prescriptions – most EHR systems let physicians cut and paste notes from previous visits or other sources. This can be useful when writing referral letters or admission notes, but can also lead to redundancies and copying of information that is no longer true or relevant.
Long-term solutions for these challenges are still evolving. The current standard for data-entry is a semi-structured approach, combining narrative text fields for NLP with structured data entry. There is an additional challenge of transferring data recorded prior to EHR installation, which sees a number of approaches implemented. Some organizations connect EHR interfaces to electronic sources and load data 6-12 months before the system goes live. Some extract selected data and hand enter it into the EHR before the patient's first visit after the system has been installed. Others scan and store 1-2 years of old paper records using novel technology for printed text reading.
2.1.2 Data Validation and Visualization
With by-hand data entry comes a high chance of transcription errors, and thus a need for discriminating validity checks. There are five commonly-used validity checks to ensure input data is accurate and consistent: range checks; pattern checks; computed checks; consistency checks; delta checks; and spelling checks. Range checks detect or prevent entry values out of range (e.g. entering 0000 or 9999 in place of a year). Pattern checks verify that data have a required pattern, such as an area code, or six digits for a zipcode. Computer checks look for mathematical relationships, such as white blood cell differentials adding up to 100. Consistency checks compare data to detect errors. Similarly, delta checks look at new and old results or observations to alert users of unlikely differences. Spelling checks, as one might expect, confirm correct spelling.
Data housed in EHRs can be presented in numerous formats depending on the context or desired use, many of which are novel with no counterpart in manual systems. In order to reap the many advantages in aggregating data and long-term visualization, computers and systems that can process data to produce these summary reports are required.
2.1.3 Data Search, Query, and Surveillance
It is a previously-established challenge for physicians to manually search a patient's chart for specific information, even before the advent of electronic systems. The underlying issue is that many of the questions clinicians ask cannot always be answered by simply scanning a patient's record – paper based or electronic. These questions can range from whether a specific test as been performed to medications that have been previously prescribed or how a patient has responded to prior treatments. Thus, the need for search tools to locate relevant data is both persistent and not yet sufficiently addressed. A key issue in EHR development remains the ability to display data in presentation formats that allow clinicians to draw important conclusions.
Query and surveillance systems, by contrast, have no manual record counterpart. These tools can be harnessed by clinicians to analyze patient outcomes and practice patterns, or surveil disease emergence based on reporting functions. While query surveillance does not directly align with clinical decision support, the underlying logic is similar – accessing records that satisfy given criteria and exporting the selected data.
2.2 Enablers and Disablers
Several areas of EHR system development and access present persistent challenges that, sufficiently addressed, enable successful development, implementation, use, funding, and policy support.
2.2.1 Information Needs
As previously mentioned, some organizations require healthcare providers to enter in all patient data generated – orders, notes, prescriptions, etc. – directly into the EHR with the goal of augmented operational efficiency. However, as we also discussed, this system puts the responsibility and time costs solely on the physicians, who in their daily practice may lose efficiency. EHRs must strike a balance between organizational and physician-level interests. System developers, then, need to understand the information needs and workflows applied in all settings where healthcare is delivered.
Successful systems have been developed either by clinicians themselves or in close collaboration with practicing healthcare providers. The questions we discussed earlier that are difficult to answer from paper charts, but commonly asked by physicians, require some synthesis of information. Most clinical systems do not have the capability to easily do so. Thus, developers must have a thorough grasp of these needs and workflows to both balance stakeholder interests and produce systems that answer questions and support decisions.
2.2.2 Usability
A key part of successful and useful systems is an intuitive and efficient user interface, which requires understanding of complex human-computer interactions and professional clinical workflows such that systems can be built that are easy to learn and use. An additional level of complexity is added with EHRs, where different users require different interfaces, between clinicians entering patient data, clerks entering patient charges, and patients accessing records, test results, or charges.
Generally speaking, users – especially clinicians – want fast response times and as few data input fields as possibles. For clinicians, any input field must have all possible synonyms across all possible users loaded into the underlying menu or vocabulary. Time conservation also requires keyboard options for all inputs. Developers of these interfaces need to focus on the unique information needs that apply to EHR systems, including information needs and tasks that will be performed. Together, these will influence information presentation.
2.2.3 Standards
National standards are critical to the development, implementation, and use of EHR systems. They reduce development costs, increase integration, facilitate collection of meaningful aggregate data, and promote both quality improvement and policy development. In developing standards, particular attention should be given to ensure that health information remains consistent across providers that a single patient may access over the course of their care. From this idea we have uniform standards, which exist to enable interoperability and meaningful data exchange. Privacy and security standards (e.g. HIPAA legislation) are also crucial for protecting individually identifiable health data.
3. Evaluation of EHR Systems
In 1993, the DeLone research group published their DeLone and McLean Model of Information System (D&M IS) to model the information success based on the work of Shannon and Weavor in the 1980s. Ten years later, they followed up with a systematic review about their successfully and widely applied D&M IS model to refine or update the old definitions and structures, and suggest for new variables for consideration (Delone, 2003). Their original theory model had a framework consisting of six independent variables: system quality, information quality, use, user satisfaction, individual impact, and organizational impact, while the universal dependent variable was the IS success. Here, “Information quality” concerns whether the data in the EHR are relevant, comprehensive, precise, and provide an adequate overview of clinical work. “System quality” addresses whether a system has the required functionality to support the work in question, and involves issues such as functionality, integration with other systems, performance, stability, and ease of use. “Service quality” addresses the support available to users of a system. User training was included in this evaluation, as it was regarded as possibly providing essential information for future implementations. They are the critical dimensions for systematic performance.
Figure 1. The Updated Delone and McLean Framework for Evaluating Information Systems Success
Successful implementation of EHRs generally requires satisfaction on the part of the staff (Vikkelsø, 2005). In 2013, the Bossen research group conducted an evaluation of the EHR systems in Denmark based on the theoretical DeLone McLean model. They chose to use the model because it had been validated by other studies, and combined a comprehensive scope, with regard to the different aspects that may influence information system success, and a relatively simple model, using only six dimensions (Bossen, 2013). They collected feedback from physicians, nurses, medical secretaries, and physiotherapists in forms of questionnaires, semi-structured interviews, and focus group interviews before and after the implementation of the EHR system. They also included the ethnographic observation and performance data from the four groups. The researchers indicated that the participants preferred the EHR system to be user-friendly, quickly responding, well-structured with dictation, well-established with an overview of patient information, timely updated, with appropriate data accessibility (Bossen, 2013). They also indicated that a successful EHR system should be pre-implemented with optimization of hardware and network, collaborating with a qualified in-house implementation team and management support staff, and demonstrating a high degree of user involvement and flexible configuration (Bossen, 2013). Meanwhile, suspension of productivity demands and the need for success ensures cooperation between central actors. And their insights echoed with what we summarized in 2.2.
4.EHR-driven Research and Applications
4.1 EHR Biobanks Overcome the Limitations of Traditional Randomized Control Trials
The traditional approach to evaluating the relationships between disease risk factors and intervention or treatment are randomized control trials (RCT), recruiting participants in cohort or case-control settings. In general, they could reflect sufficient evidence to support a given medical intervention (Nair, 2016), yet they may suffer from the dilemma of the realistic cost of time and budget for recruitment, the study design limitation, and the selection bias of sampling (Nair, 2016). On the other hand, the RCT with a limited sample size may not have large enough statistical power or reduce the possibility of capturing statistical significance in R.F.s (Verma, 2018). With the wide implementation of electronic health record (EHR) systems all over the U.S., the aforementioned limitations could be overcome to an intermediate extent or above. EHR could capture much larger amounts and types of patient information than the pre-selected RCT variables, including yet not limited to social and behavioral factors(e.g., socio-economic determinants, tobacco and alcohol consumption) with satisfactory sample sizes (Adler, 2015) Meanwhile, EHR could serve for secondary analysis with coding unstructured clinical narratives with aid from natural language processing algorithms and clinical standards such as International Classification of Diseases 9 (ICD-9), International Classification of Diseases-10 (ICD-10), and Health Level-7 (HL-7) (Roberts, 2015; Denny, 2013; Stanaway, 2019) The variables generated from EHR may contribute to the progression of diseases of interest and help scientists for designing interventions.
The EHR has been proven to enable a vision of “personalized medicine” in which genomic and other information are incorporated into precision medicine (McCarty, 2011). Studies using clinical measurements from electronic health records (EHRs) permit a long-term average of multiple independent clinical measurements from many different clinical visits, yielding reduced phenotypic variance (Ganesh 2014). Accompanied with appropriate quality control and coding granularity during integration from various data providers maintaining, the scope of pooled EHR, ranging from tens of thousands to millions of samples, afford insight into treatment or disease effects that may be under-reported or missed in the underpowered RCTs (Nair, 2016). Many biobanks are beginning to link data from study participants’ or patients’ EHR with their genetic data to accelerate the study of genotype-phenotype associations, while potentially facilitating clinical activities in the absence of genetic specialists (Lemke, 2010; Ayatollahi, 2019). In late 2007, the National Human Genome Research Institute (NHGRI) established the Electronic MEdical Records and GEnomics (eMERGE) Consortium (www.gwas.net) to develop, disseminate, and apply approaches to research that combine DNA biorepositories with EHR for large-scale, high-throughput genetic research (McCarty, 2011). Now, the consortium has expanded to nine study sites, including the University of Washington and Group Health Cooperative in Washington, and reached its third phase. Beginning from its first phase, eMERGE has conducted studies that successfully used EHR data to classify case or control, generate standardized disease-related phenotypes, and impute genotypes for relevant genome-wide association studies (GWAS) (Gottesman, 2013; Kho, 2011). The eMERGE allowed researchers to conduct powerful GWAS, which required more than 10000 samples, which were hard to reach during traditional RCTs (Park, 2010). The data mining on unstructured portions of EHR also suggests disease phenotypes, which could be used as responses of GWAS (Bush, 2016).
4.2 EHR-driven Analysis Could Suggest Preventive Interventions for Public Health
EHR data has been used for surveillance of clinical disease risk factors (e.g., CVD and its RF such as body mass index, blood pressure, and cholesterol levels) for population health improvement initiatives (Sidebottom, 2014). EHR could also enrich the information necessary for clinical decision making by reminding higher-risk populations for regular preventive screening (e.g., cancer screening). Cancer development originates from the uncontrolled proliferation of healthy cells due to genetic mutation caused by UV exposure, and/or contact with physical or chemical carcinogens. The expansion of cancer cells could possibly be prevented before the benign cells enter the stage of metastasis. Although the prediction of cancer from EHR is not quite approachable, it was suggested that community-based cancer screening could hugely raise the measurable cost-effectiveness of treatment or prevention for cancer progression (Goldie, 2005). If residents participate in regular screening for cancers, the healthcare providers could capture more true positives of cancer incidences at an earlier stage of cancer. Carcinogenesis may proceed much more slowly with early intervention or treatment to prevent enormous costs for late stages. Meanwhile, the awareness of cancer examination with aid from educational materials from public health stakeholders could guide residents with a healthier lifestyle for cancer prevention. Early detection and intervention driven by EHR-based surveillance will likely lower the cost of surgery and treatment. EHR has also been used for revealing and characterizing the disease factors among the linked open data for inference and prediction(Boytcheva, 2019). By analyzing the patient data on a large scale, we may find out more cancer biomarkers and causal factors for the optimization of screening protocols and excellent comprehension of carcinogenesis for treatment and prevention. Observational epidemiology studies have found that CVD-/lifestyle-related RFs are associated with dementia risk and targeting these modifiable RFs (e.g., diabetes, hypertension, obesity, physical inactivity) could present a viable dementia prevention strategy which prevented 35% of dementia (Barnes, 2011; Norton, 2014; Broce, 2019).
Reference:
Bossen, C., Jensen, L., & Udsen, F. (2013). Evaluation of a comprehensive EHR based on the DeLone and McLean model for IS success: Approach, results, and success factors. International Journal Of Medical Informatics, 82(10), 940-953. https://doi.org/10.1016/j.ijmedinf.2013.05.010
Clinical Decision Support. (2018, April 10). Retrieved April 19, 2020, from https://www.healthit.gov/topic/safety/clinical-decision-support
W.H. DeLone, E.R. McLean, The DeLone and McLean model of information systems success: a ten-year update, JMIS 19 (4) (2003) 9–30.
Fossey, R., Kochan, D., Winkler, E., Pacyna, J., Olson, J., & Thibodeau, S. et al. (2018). Ethical Considerations Related to Return of Results from Genomic Medicine Projects: The eMERGE Network (Phase III) Experience. Journal Of Personalized Medicine, 8(1), 2. https://doi.org/10.3390/jpm8010002
Ganesh, S. K., Chasman, D. I., Larson, M. G., Guo, X., Verwoert, G., Bis, J. C., Gu, X., Smith, A. V., Yang, M. L., Zhang, Y., Ehret, G., Rose, L. M., Hwang, S. J., Papanicolau, G. J., Sijbrands, E. J., Rice, K., Eiriksdottir, G., Pihur, V., Ridker, P. M., Vasan, R. S., … Chakravarti, A. (2014). Effects of long-term averaging of quantitative blood pressure traits on the detection of genetic associations. American journal of human genetics, 95(1), 49–65. https://doi.org/10.1016/j.ajhg.2014.06.002
Glessner, J., Li, J., Desai, A., Palmer, M., Kim, D., & Lucas, A. et al. (2020). CNV Association of Diverse Clinical Phenotypes from eMERGE reveals novel disease biology underlying cardiovascular disease. International Journal Of Cardiology, 298, 107-113. https://doi.org/10.1016/j.ijcard.2019.07.058
Goehringer, J. M., Bonhag, M. A., Jones, L. K., Schmidlen, T., Schwartz, M., Rahm, A. K., Williams, J. L., & Williams, M. S. (2018). Generation and Implementation of a Patient-Centered and Patient-Facing Genomic Test Report in the EHR. EGEMS (Washington, DC), 6(1), 14. https://doi.org/10.5334/egems.256
Goldie, S.; Gaffikin, L.; Goldhaber-Fiebert, J.; Gordillo-Tobar, A.; Levin, C.; Mahé, C.; Wright, T. Cost-Effectiveness Of Cervical-Cancer Screening In Five Developing Countries. New England Journal of Medicine 2005, 353, 2158-2168.
Muse, E., Wineinger, N., Schrader, B., Molparia, B., Spencer, E., & Bodian, D. et al. (2017). Moving Beyond Clinical Risk Scores with a Mobile App for the Genomic Risk of Coronary Artery Disease. https://doi.org/10.1101/101519
Nair, S., Hsu, D., & Celi, L. (2016). Challenges and Opportunities in Secondary Analyses of Electronic Health Record Data. Secondary Analysis Of Electronic Health Records, 17-26. https://doi.org/10.1007/978-3-319-43742-2_3
Norton S, Matthews FE, Barnes DE, Yaffe K, Brayne C (2014) Potential for primary prevention of Alzheimer’s disease: an analysis of population-based data. Lancet Neurol 13(8):788–794. 10.1016/S1474-4422(14)70136-X
Ott, J., Wang, J., & Leal, S. (2015). Genetic linkage analysis in the age of whole-genome sequencing. Nature Reviews Genetics, 16(5), 275-284. https://doi.org/10.1038/nrg3908 Park, J., Wacholder, S., Gail, M., Peters, U., Jacobs, K., Chanock, S., & Chatterjee, N. (2010). Estimation of effect size distribution from genome-wide association studies and implications for future discoveries. Nature Genetics,42(7), 570-575. https://doi.org/10.1038/ng.610
J. E. Petrikin, L. K. Willig, L. D. Smith, S. F. Kingsmore, Rapid whole genome sequencing and precision neonatology. Semin. Perinatol. 39, 623–631 (2015).
Shortliffe, E. H., & Cimino, J. J. (2014). Biomedical informatics: Computer applications in health care and biomedicine. London: Springer.
S. Vikkelsø, Subtle redistribution of work, attention and risks: electronic patient records and organisational consequences, Scand. J. Inf. Syst. 17 (1) (2005) 3–30.
What is computerized provider order entry? (2018, March 21). Retrieved April 19, 2020, from https://www.healthit.gov/faq/what-computerized-provider-order-entry Computerized Provider Order Entry. (n.d.). Retrieved April 19, 2020, from https://psnet.ahrq.gov/primer/computerized-provider-order-entry