Capstone project

Integrating Disparate Data in the Electronic Health Record to Improve Patient Care

for

Information and Communication Technology, Master of Applied Science

Project Management

UCOL Student

University of Denver University College

Month 1, 2016

Faculty: Paul Novak, M.A.

Director: Thomas J. Tierney, Ph.D.

Dean: Michael McGuire, M.L.S

Abstract

Electronic Health Records (EHR) can further integrate data and leverage it for decision making to improve clinical outcomes. Barriers have arisen due to a rapid implementation of EHRs spurred on by government incentive payments, the characteristics of data, ineffective adoption practices, and resource constraints. This paper explores the barriers to incorporating disparate data sets into an EHR, the means to address the problem, establishes best practices, and sets further recommendations to pursue. A diverse set of secondary research sources were analyzed to determine an applicable solution. This paper argues that incorporating disparate data sets into the EHR is not out of reach and should be pursued to improve clinical outcomes derived from leveraging data in the EHR.

Table of Contents

Background 1 Competing Views 1 EHRs Lack Data 2 Why More Data Matters 2 Decisions Improve with More Data 2 More Data Leads to Empowerment 3 Approach 3 Discovery and Creation of an Argument 3 Types of Information Used 4 Peer-Reviewed Journal Articles 4 Research Based Books 4 Articles from Industry Leaders 5 Government Publications 5 Case Studies 5 Industry Communities and Newsletters 5 Forming Solutions 6 Findings Led to Further Research 6 Findings Drove the Creation of Solutions 6 Literature Review 7 EHR Data Characteristics that Create Barriers 7 Data is Stored Differently 7 Unstructured Data is Common 8 Methods to Manage Unstructured Data 9 Meaningful Use Criteria 10 Outside Factors that Create Barriers 11 Consequences of the HITECH Act and Meaningful Use 11 Time and Cost Concerns 12 Poor Adoption Undertakings 12 Relationship Between Information 13 Solution 15 Improved Adoption 15 Modify Policy 16 Support Management of Unstructured Data 17 Budget and Cut Waste 17 Best Practices 18 Engage Leadership 18 Prioritize Projects 18 Negotiate for Data Incorporation and Costs 19 Accept a Long-Term Outlook 19 Discussion 20 Strengths 21 Weaknesses 22 Opportunities for Improvement 23 Environmental Threats 24 Recommendations 24 Conclusion 25 References 27

Student-v

The implementation and use of Electronic Health Records (EHR) has steadily risen over the last few decades and is becoming increasingly important in providing healthcare as clinicians move away from paper based records and into electronic formats. A survey conducted by the National Ambulatory Medical Care Survey (NAMCS) in 2013 showed that 78% of office-based physicians used some type of EHR, which is a significant increase from the 18% of office-based physicians using an EHR in 2001 (Center for Disease Control and Prevention 2014). The pass of the Health Information Technology for Economic and Clinical Health (HITECH) Act in 2009, which provides incentive payments to organizations using an EHR as long as meaningful use criteria is met, is a strong driving force behind the rapid implementation of EHRs (Center for Disease Control and Prevention 2014). Meaningful use criteria defines specific outcomes that recipients must attest to in three different stages over the next five years, such as securing data in a specific way and using it to track key conditions (healthit.gov 2015). Yet healthcare organizations have yet to expand the data contained in EHRs to disparate systems and incorporate as wide of a data set as possible.

Incorporating data from separate applications, even applications still managed by the same organization, is controversial because of the challenges faced and costly solutions; the need to do so is far from agreed upon. Multiple approaches exist including middleware, expanding EHRs, building interfaces that bridge applications, or forfeiting the incorporation of outside data and foregoing the associated costs and use of resources. Should an organization want to expand their EHR to include more data for decision making, regardless of which decision is made the cost is often hundreds of thousands, if not millions of dollars due to the required time of skilled workers and complexity of work.

A national survey showed that out of nearly 200 organizations surveyed, 68 responded that they needed to build 10 or more time consuming and costly interfaces to support the sharing of data between systems (Commins 2014). Organizations that cannot commit the required resources to build such interfaces are unable to incorporate available data into the EHR. Furthermore, a barrier is created when organizations intentionally do not share data because of concerns that data or the EHR is proprietary and a competitive advantage (Commins 2014). Beyond the existing data that can be included, emerging technologies such as wearable fitness devices and technologies not yet discovered have a potential to provide relevant data. This data can describe a patient’s health to their clinician if included in an EHR, but data from these devices has not been supported by EHRs due to their recent arrival and inability for EHRs to keep pace with vast emerging technologies.

Entire applications have been built to improve decision making in the clinical environment by leveraging data. This has gone to the extent that new business markets have developed for such products and services. The most notable of these applications are computerized clinical decision support systems (CDSS) which present useful data to healthcare providers, and are well inside the capability and inclusion of EHR offerings. Leveraging such data improves clinical outcomes such as patient safety, cost reductions, reduction in adverse drug events (ADE), and an overall improved quality of care (Beeler 2014). The impact of these systems is tremendous, a survey of six community hospitals near Boston showed that 9 out of 10 ADEs were preventable and could likely have been avoided by a CDS system that featured renal dose checking (Beeler 2014). The more data that is stored and manipulated into meaningful information for use in decision making applications, then the greater the ability for more informed decisions.

Integrating data that is documented or generated from a patient empowers them to manage their own health (Blumenthal 2014). A more active role by patients in managing their health could reduce healthcare costs and provide a better quality of life resulting from the proactive ability of the patient detect medical conditions early. Every individual is reliant on his or her health and as such the care provided, providing credence that an improvement to healthcare should not be relinquished but pursued.

The argument explored is that implementing the solution established herein improves standardization and adoption processes, policies, and systems to allow data that is stored differently across applications to be incorporated in EHRs, creating a comprehensive set of information which will improve healthcare decision making and consequently clinical outcomes.

The problem that data is inadequately integrated in EHRs was defined from experiences in the healthcare and technology industries. In order to gain a full understanding and further define the problem, various types of sources were reviewed for credibility and used if appropriate. All research incorporated in this paper is secondary in nature and no primary research was conducted or used during any point of analysis or research. Using secondary research, the problem was further defined and clearly established which led to a set of anticipated solutions. Additional research was conducted to compare and contrast best practices in areas that need improvement. An analysis of research generated a formal set of solutions and recommendations that can be used to improve the outcomes of EHR projects and data integration. Throughout all phases of research, specific attention was given to use the most current information possible due to the rapid advancements in health information technology.

Peer-reviewed journal articles provide a cornerstone for research due to abundance of such information in the medical community. Such articles serve as a direct source from healthcare professionals and provide a unique perspective.

Books provide for an easy reference and traditional source of information. Only books based on research were utilized, most notably Beyond Implementation authored by Dr. Haugen and Dr. Woodside. This book is of particular use as a strong base of research served as the foundation of the content and the authors have multiple accolades. Furthermore, Dr. Haugen has over 15 years of research experience and attained a doctorate in health information technology from the University of Colorado Health Sciences Center (Haugen and Woodside 2010). Dr. Woodside has over 30 years of experience and has served in various roles including executive vice president and chief medical officer at UT Medical Group, Inc. (Haugen and Woodside 2010).

Articles from medical associations such as AHIMA are incorporated in this research to provide insights from industry leaders that drive aspects of healthcare. These organizations are well established and often provide direction to healthcare organization or set requirements for health programs.

Information attained from surveys conducted by the government provides a broad spectrum for research and ample data for analysis. These sources are often unique as the government undertakes such studies to assess policy and one often singular in the ability to conduct extensive statistical studies on a national scale.

While case studies are difficult to find, they provide a unique ability to directly describe the story of data for healthcare organizations and provide a hospitals perspective. As a result, case studies are an excellent source to identify the struggles that healthcare organizations face, as well as the value that has been gained or lost by changes to processes, systems, and data that is leveraged in EHRs.

These sources were not used to provide hard facts or established solutions because concerns arise surrounding the possibilities of subjective information or inaccurate reporting. Instead, these sources served as a way to identify trends, ideas, struggles, and demands that are voiced by the online communities. Current information is readily available which makes newsletters an effective means to determine what hot topics are related to problems and solutions. However, KDnuggets is seen as credible due to its editor’s recognition as an expert in the industry, and accolades including Forbes naming the organization as a top influencer in its industry (KDnuggets 2015).

Analysis of research led to related and opposing views to be sought in further research. Conducting research that encompassed multiple views established what information is controversial, and provided counter arguments that must be considered. This was an intentional processes used as a way to aide in an objective evaluation of material that avoids cognitive biases. Specific care was given to avoid framing bias which leads an audience to a conclusion dependent on the way facts are presented, and the primacy effect which leads a reader to recall the initial set of information first (Reich 2013).

Following the findings of research resulted in a process that drove the creation of solutions. After the literature review was completed, an effort to seek out and identify best practices was made in order to provide solutions and support applicable recommendations. It is imperative that best practices were not sought at out early, as this would direct research in a manner that supports a solution that may not be objective in nature. Once solutions have been established, the counter arguments were considered to determine the most appropriate recommendations that address conflicting information and controversial ideas.

Data is not organized and stored in the same manner across applications which make inclusion of further data difficult. Relational databases are by far the most common form of databases in healthcare applications (Campbell 2004). Databases are the foundation of applications because they store data that is ultimately presented and used, but changing databases can be expensive. As a result, it is difficult to transition to a different database management system often referred to as a DBMS. Specifically, relational databases organize data in rows and columns, where each column contains a specific type of value, and values must be of the same format (Dictionary.com). How is this a problem? Applications organize relational databases differently, and what serves as an identifier in one database is not the same as another database. As an example, a Social Security Number SSN can be used as a unique identifier in one database to identify a patient, but a separate database used by a different application could use a generated identifier from the patient’s name and date of birth. So how can one application reference the data contained in a database from a disparate application or different data source? An identifier must be established between what is present in both sets of data, or must be able to be generated in both sets of data. This creates limitations in how data can be referenced due to a gap in standard design between applications. Until standards exist across applications to identify unique individuals, doing so will be unrealized (Tripathi 2012). Further research continues to build upon this notion. When considering the challenges of integrating genomic data, additional standards need to be established for data so that improvements to integration can be made possible (Connolly, et al. 2013). Although this identifies genomic data, the analysis can be applied to all data as a lack of standards is indiscriminate as to what the data depicts, only how it is organized and stored. Analysis of the data cannot be misinterpreted and clearly identifies the lack of consistent standards as a significant barrier to incorporating further data. Further data incorporation is not impossible but increases costs, complexity, and time required to integrate additional data sets. How does this impact an EHR?

Data to integrate is often unstructured or free text. Paper forms incorporated into EHRs contain large amounts of unstructured data in the form of free text. Paper forms can either be scanned in as a document or entered into an EHR by a data entry clerk. Patient generated notes are impeded for not only the same reasons, but also because a patient is unable to structure their notes and logs in a manner that easily transfers to a relational database because these notes are not in rows or columns and lack structure. Recent products such as Fitbit devices that monitor heart rates cannot readily transmit or incorporate data into EHRs at this point because data is stored differently and patients do not have an easy or effective way to retrieve and incorporate the data into their EHR. Furthermore, EHRs typically include fields for free text and effectively acquiring this unstructured data limits the ability for free text such as progress notes, admission notes, or general free text (Johnson et al. 2008). Johnson (2014) continues to note that physicians often prefer to use a narrative based approach using free text because it is quicker than entering and capturing the data in a highly structured form that becomes time consuming as the scope or variation departs from standard workflow events.

A further exploration shows that new technologies are emerging that can effectively manage and query unstructured data that traditional databases cannot; the most notable solution is Hadoop. The outlook of Hadoop and similar management software is very promising. Experts anticipate that this technology will be applied far into the future, with some stating that they will replace the traditional relational databases that are widely used in corporations and healthcare organizations (Sanghavi 2014). Hadoop and similar software are able to group and store data together in a “pool” that permits the retrieval of data as long as it can be queried. This greatly increases scalability because the data being stored is no longer required to be in a specific format and does not undergo Extraction, Transformation, and Loading (ETL) processes, so new sets of data can be stored more quickly (CITO Research 2014). ETL is a process that brings data from different data sources, transforms the data into a single standard format, and then stores it in a structured manner. Research does vary on this outlook, as is the case in most areas of technology. According to a study by Gartner in 2014, analysts and data administrators using Hadoop and similar tools to manage unstructured data must be highly skilled at manipulating and analyzing data. The need for such skill arises because unstructured data that is incorporated into these technologies remains in its native format and does not undergo ETL which causes a lack of consistency when referencing or querying data (Gartner 2014). The largest problem may be that each query is custom built and one cannot easily determine how another generated their information from the data, resulting in the possibilities of two different versions of what is true for one set of data (Gartner 2014). This reduces credibility of reports generated from data and underlines the strong need for those analyzing the data to be intimately familiar with the data sets to avoid generating two sets of information from the same data. These approaches do not remove the problem of standardization, but approach it by storing and referencing data differently.

A new method, structural narrative, has yet to be employed in EHRs other than at an experimental level, but has the potential to mitigate problems of standardization to allow further incorporation of data. Structural narrative allows physicians to enter free text paragraphs inside of standard fields, however, the free text paragraphs are marked up using Extensible Markup Language (XML) (Johnson 2014). This allows a processing of the free text paragraph that identifies key words such as diseases, medications, procedures, modifiers, and other elements that can be separated and stored in a structured manner (Johnson 2014). This is possible because the restrictions imposed on where free text can be entered allows it to be limited in its applicable context because it will relate to the standard field selected as the physician moves through the workflow. To use this method, a database system must be employed that enables XML storage or the data must be transformed and stored in a different format standard. Johnson (2014) further states this method presents risk that free text could be identified incorrectly, though additional constraints can be implemented and further evaluation is required to determine error rates. Because this method is experimental, clinician opinions and field results are unavailable.

The passage of the Health Information Technology for Economic and Clinical Health (HITECH) in 2009 accelerated the rate of adoption of EHRS by providing incentive payments for hospitals (Tripathi 2012). These incentive payments are based on meaningful use requirements. Importantly, the meaningful use standards dictate that data must be captured in a standardized format and how it must be shared (healthit.gov 2015). However, the requirements for standards are not granular in nature, so organizations can vary how the database stores data. Providing such granular requirements would greatly limit the design of databases to facilitate the creation of a standard. Furthermore, meaningful use requirements for the final stage are not yet decided and are subject to change (healthit.gov 2015). As a result, EHRs face a challenge in ensuring that they are most aligned with the requirements of the final stage. In addition, healthcare organizations are unable to establish the details of long-term planning in workflows, how the EHR must operate, or what system architecture and development is required to meet their needs. This is important because functionality can drive design of systems and applications, which is largely dependent on a database and its structure. Standardization cannot be fully achieved at this point as evidenced by ongoing proposals of meaningful use criteria that serve as a barrier. Beyond the current data that exists, new technology and forms of data will emerge.

The rapid adoption of EHRs has spurred hospitals to create them without adequate roadmaps and has contributed to the creation of disparate systems with separate data sets (Studeny and Coustasse 2014). Shifts in meaningful use criteria and the effort to qualify for incentive payments has further contributed to a lack of integration and poor interoperability between systems; the incentives support creation of data silos (Studeny and Coustasse 2014). In order for a healthcare organization to incorporate data, system architecture must first be implemented in a manner that supports their needs and is managed to enable quick integration of new data. As evidenced by Studeny and Coustasse (2014), the establishment of such environments is outpaced by the established timeframes to achieve incentive payments and for some, avoid financial penalties.

Implementing an EHR can vary widely in cost, but estimates range from $15,000 to $70,000 per provider (healthit.gov 2015). Costs can vary widely based off the EHR vendor and what EHR functions or components are included. A hospital with 100 providers could possibly spend tens of millions of dollars, which may exhaust a budget. Currently if more data sets are to be incorporated, each set of data from a disparate application linked to an EHR or incorporated in a separate database must undergo the ETL process. ETL is a complicated, expensive, and time consuming process (Mai 2014). ETL is costly in part because software licenses can be expensive and highly skilled workers are needed to oversee the process to ensure the quality of data. Healthcare is particularly vulnerable because data is used to make life and death decisions. ETL only addresses the process of standardizing the data into a structured format in a new location, but does not include effort to use the data. The development of tasks defining how data is used, such as developing a reporting system, must be considered in addition to the cost of ETL. Moreover, system architecture may need upgraded to adapt to the change in data structure, and provide more storage if data is stored in multiple places in the event of a system failure.

Not all barriers to improving the volume and use of data are technical in nature. The adoption and effective use of an EHR will affect the clinical outcomes derived from it and determine its value. If an EHR is not valuable, it risks being seen as a hindrance and support lacks to incorporate further data. True effective adoption of EHRs has been impeded by inadequate education from traditional training methods that do not provide effective delivery of education (Haugen and Woodside 2010). As meaningful use criteria changes and EHRs evolve with new capabilities, new workflows emerge and a need for further training becomes apparent. Furthermore, adoption is often hindered by a lack of engagement on the part of leadership that contributes to resistance, particularly when physicians are not involved at the outset (Haugen and Woodside 2010). Resistance to adoption can be mitigated by using an alternative approach that moves away from lengthy classroom sessions and enables online simulation training accompanied by engaged leadership, metrics, and ongoing sustainment efforts (Haugen and Woodside 2010). It is pertinent to note that such a solution also supports clinicians in reaching proficiency faster and providing more time back for patient care hours and less time spent in rigid classroom training sessions (Haugen and Woodside 2010). Incorporating new data brings changes to workflows and features of EHRs that providers must adopt. Reducing the time it takes a clinician to adopt these changes in EHRs, lessens stress and reduces the overwhelming sense of the process, which contributes to lowering resistance to EHRs and increasing the value and support to utilize data further.

Overall, research suggests that the barriers to improving the use and sources of data in EHRs are interrelated. EHR implementation has been nearly exponential; a 2013 national survey showed that in Minnesota 94% of physicians reported using some type of EHR (Center for Disease Control and Prevention 2014). This high usage underscores the importance of EHRs due to the widespread use, and thus the data that is leveraged by EHRs and the applications within them. The pace of adoption has been rapid as evidenced by the 336% increase in EHR implementation from 2006 to 2013, which has reverberating affects (Center for Disease Control and Prevention 2014). Problems with standardization have arisen because EHRs have quickly been adopted before standards can emerge or mature. This directly relates to the data, which must be well organized in a standard format, and as such requires standards. On one hand, meaningful use and the HITECH Act have spurred rapid adoption of EHRs that lack interoperability and have caused organizations to develop disparate applications. On the other hand, meaningful use is working to establish standards for how data is captured and used in EHRs which can lead to improved standards in healthcare in future years. The passage of the HITECH Act also relates to adoption and education as clinicians are thrust into using a new system and application.

Adoption is important because it provides value in the use of the EHR and data, which supports further expansion of more data. Furthermore, effective adoption comes full circle by supporting the fundamental goal of incorporating more data; it improves the clinical outcomes for patients and healthcare providers.

The high cost of formatting data into a relational database from one application to another suggests that alternative solutions can be more cost effective. Because ETL is used to provide standardization among the data from disparate applications, alternative solutions must consider the impact on addressing the characteristics of free text and unstructured data. Research suggests that alternatives to using highly structured data, such as Hadoop software or experimental processes like structured narrative, each bring their own considerations. These opposing considerations must be balanced against one another when determining actionable next steps. Furthermore, time and cost concerns relate to the adoption undertaking and how far education is pursued based off the allotted resources of time and money.

Meaningful use must be considered in all stages as organizations strive to attain incentive payments. Poor adoption will contribute to a lack of improved clinical outcomes and a failure to reach the criteria established. According to Healthit.gov (2015), unstructured data and free text must be captured in a manner approved by meaningful use, and perhaps most importantly meaningful use itself must be considered. Criteria are not yet approved for the final stage and as a result long-term planning is difficult which impacts all barriers identified because each barrier requires considerable planning when forming a solution.

The solution requires a multipronged approach. Each barrier must be addressed as part of a comprehensive solution; no single action can readily permit further data integration into EHRs to improve clinical outcomes. All components of the solution culminate to an effective means when addressing barriers for large hospitals or integrated healthcare networks. While the solution is not as applicable for smaller hospitals, certain components are beneficial for their use, such as adoption solutions and policy reviews which can be pursued to some benefit.

Improved adoption provides greater value to the data, and resistance to further data incorporation can be mitigated by improving the adoption solutions offered to clinicians. Using an online approach is ideal to reduce hours spent in classroom training and provide clinicians the flexibility and empowerment desired when learning new tasks and major changes within an EHR. Improved adoption techniques such as online simulations allow a clinician to quickly gain proficient use of EHRs which is imperative so they are ready for further use of data and value is realized from the inclusions of data. Leadership must be engaged in the process to reduce resistance to change and adoption, as well as provide proper governance from the top down (Haugen and Woodside 2010). Adoption must be viewed as a long term endeavor and not short-term education, because sustaining adoption over time will yield better use of the EHR.

Hospital organizations have little control over policy implemented for meaningful use other than making suggestions. Hospital organizations need to make their concerns heard, appeal for more standards, and most importantly request an extended timeline. Internal policies can be reviewed to ensure that the hospital is not striving to achieve incentive payments at the cost of poor long-term planning and substandard system and database design. The criteria should further standardize data manipulation and restrict the use of free text. Incentives for Interoperability need to be extended because findings show a lack of interoperability that has been caused by a rush to implement EHRs and Personal Health Records due to financial incentives (Studeny and Coustasse 2014). Delaying the current deadlines for hospitals to qualify for incentive payments will provide healthcare organizations time to properly plan and incorporate more data in their EHRs. It is unreasonable to expect hospitals to easily respond to a change that has driven EHR implementation to an increased rate of 336% in the years spanning from 2006 to 2013 (Center for Disease Control and Prevention 2014). This will support organizations in developing roadmaps and integration points for data between applications that otherwise would be left to the wayside in an attempt to meet requirements by an established date and avoid financial penalties.

Solutions such as Hadoop have emerged in recent years that allow unstructured data to be queried into reports. EHR vendors should support the use of unstructured data in future releases where applicable, without providing such a great use of free text that reports are heavily based off of the unstructured data due to concerns of accuracy in relation to queries. In the event EHR vendors are unresponsive to this need, organizations should work with vendors at the contractual level to jointly provide support for unstructured data. Such a solution should only be used when data cannot be captured in a structured manner and used until an organization has the capability to structure the data into a manner consistent with the standards of the EHR database. Alternatively, structured narrative can be implemented in specific, low risk workflows that have a minimal scope and variation.

Hospitals must budget for increased expenses associated with EHRs so further data can be incorporated and cost savings realized in the long-term. In order to increase the available budget, expenses can be reduced by reviewing internal policies to identify waste. Projects need to be prioritized in part on what financial return they will provide to the hospital, which is a shift for some organizations, and incorporating data should reside as a leading priority. Hospital organizations need to maximize their development efforts to qualify for incentive payments, but also to begin attaining value from incorporating and leveraging additional data early in the use of their EHR(s).

Each best practice is a low cost principle that can be utilized alongside the solution to improve the likelihood of a successful effort. Best practices are based from the research analyzed in the literature review or are considerations that influence a solution that must be kept in mind. These best practices are appropriate for any size of hospital or healthcare organization because extensive resources are not required to implement or pursue these established methods.

Research suggests that engaged leadership is a primary driver of adoption, but can serve as a barrier if leadership does not provide governance or is not involved (Haugen and Woodside 2010). In pursuing the adoption component of this solution, it is imperative that special attention is given to engage leadership in order to provide governance and set a tone from the most visible members of the organization. It is also imperative that adoption encompasses a long-term outlook and commitment, adoption is not a short endeavor or merely an implementation (Haugen and Woodside 2010).

In order to effectively reduce budget cut and waste, projects need to be prioritized properly. Best practices suggest using a weighted analysis to avoid subjective influences affecting decisions for which projects receive priority. The organizations goals need to be established and criteria for meeting those goals should serve as the units of measurement in the weighted analysis. Simplistic yes and no questions should be avoided where possible to provide a scale that employs proper weights to each element and evaluates each component in a project accurately to determine its priority. Special consideration to the cost and return on investment need to be given; a higher return yields additional funds to be reinvested into the organization in the future.

Each hospital has a chance to negotiate with their respective EHR vendor. It is important that this opportunity is taken to ask for support of a way to utilize unstructured data and incorporate it into the EHR. Likewise the negotiation process presents an opportunity for a hospital to aggressively negotiate for lower costs. As mentioned previously, cost estimates for implementing an EHR range from $15,000 to $70,000 per provider (healthit.gov 2015). Due to this high cost hospitals should inquire how EHR licensing and implementation is billed as it could cost more a physician license or if more physicians are on the payroll. An organization can then conclude whether it would be of benefit over time to employ physicians on a contract basis or hire nurse practitioners and physician assistants instead. As the Center for Disease Control and Prevention (2014) notes over 78% of office-based physicians used some type of EHR as of 2013, signaling that competition between EHR vendors is high and EHRs will strongly compete to sell to remaining hospitals as the market for new clientele contracts.

Regarding the management of data, organizations should incorporate a long-term outlook in the design of databases and make efforts to standardize the formatting of data where possible. When a large risk is present that requirements can change that will negatively impact long-term plans, then development of such plans should wait until risk is lowered.

Using a multi-pronged approach to address the problem allows organizations to determine what measures are most responsive for their environment. The solution can then be tailored to the specific needs of the organization by leveraging the components that provide the best response, while reviewing better ways to support components of the solution that lack effectiveness. This also ensures that organizations are implementing improvements and best practices in areas that could otherwise be lacking and not yet apparent yet. A comprehensive solution can be viewed as preventive and proactive for this reason, not just reactionary.

In this case, the greatest flaw in pursuing this solution is that exceptional results cannot be expected by pursuing a single component or a partial set of the components. This limits the solution to organizations that have adequate resources to commit to the pursuit of each solution. In particular, effective adoption education requires great input from clinicians and subject matter experts that often have competing priorities such as patient appointments, and clinical tasks that form part of their main job function (Haugen and Woodside 2010).

In order to further analyze the solution at a detailed level, a SWOT analysis has been conducted. The SWOT analysis in figure 1 outlines the solution’s strengths and weaknesses, areas of improvement for the solution, and environmental threats that can impact the overall effectiveness of the solution or a component of the solution. Each item of the SWOT analysis is further described in detail in the following sections to clarify.

Figure 1

Figure 1. The Solution SWOT Analysis.

The ability to experience positive results without a full success of this solution is an advantage. This strength is important because it reduces the risk that no return is provided if the effort falters in one area because other efforts may be fruitful, thus improving the changes for a positive return on investment. This is then supported by the best practice of prioritizing projects that provide a positive return.

The solution provides long-term growth because incorporating data will improve clinical decision making and ultimately reduce errors and improve care which leads to lower costs in the long-term. Improved decision making, long-term cost reductions, and long-term growth are intertwined strengths, but have separate advantages as well. Long-term growth is established by more effective use of the EHR and improved planning of data management. Clinical decision making is accomplished by incorporating further data into the system, but also by utilizing an improved approach to adoption. Cost reductions are achieved not only from leveraging more data for decision making, but also from increasing the effective use of the EHR through proper adoption.

A final but significant strength exists in that the solution has benefits outside of the EHR and can be applied to other areas. Reducing waste and low value projects from a budget will improve the overall financial standing of an organization. Improved adoption for an EHR can be established, and then replicated in other training initiatives to further reduce the hours spent training, and thus costs. Moreover, a comprehensive data set can be used in applications outside an EHR for the purposes of data mining to determine performance, accurate costs, healthcare trends from the local population, and the needs of patients.

The solution presents three main disadvantages. First, the solution is limited in its application to organizations with adequate resources to pursue each component. This narrows the scope of the solution to mostly large hospitals or integrated healthcare networks that can commit personnel to reforming adoption, reviewing policies, reducing waste in the budget, and improving available support and solutions for unstructured data.

Secondly, the solution requires multiple avenues of approach to be truly successful. Improving adoption alone will not allow for further data integration nor will a mere approach to modify policy. Acting only to improve standardization and support for unstructured data will reduce the likelihood of success. This means a significant commitment must be made by an organization to pursue the solution because it requires efforts for all components to be undertaken.

Third, the solution does not provide an immediate return of value. This can be difficult to sustain commitment and temptation to abandon the effort will mount as time passes. This is especially a weakness in organizations that do not have strong financial resources and lack adequate cash flow to support their operations or invest in project initiatives. The lack of an immediate return of value ties directly back to the first weakness as it further limits the application of the solution to audiences that can spare an investment without an immediate return.

It is important to provide an applicable solution to small organizations as well that may not have significant resources. The solution is also general in the sense that it is not highly specific to any single organization. In order to provide a solution at this level of specificity and offer the utmost chances of success it is important to determine granular details. These details would likely require a consulting service to uncover the database structure, EHR vendor in use, system architecture, available resources, expertise of staff, influencing policies, and the current education and adoption strategy. While this is an area for improvement, it is important to note that recommendations can only be specific to a certain extent as they apply to wide audience.

Additionally, the solution does not provide a quick fix. The solution takes time and while best practices can be utilized quickly, the solution requires long-term commitment. This reduces the responsiveness of the solution and limits the application to situations where immediate needs are not a constraint. This could be improved by the use of an experimental technology such as structured narrative that readily addresses the problem. However, this improvement to the solution cannot be counted on since it is only in a testing and development phase and may take years to come to fruition, if at all.

Ongoing changes in meaningful use are a significant threat to any solution at this point. If criteria changes, such as providing a specific type of access for patients with storage and reporting of data, it could nullify current efforts that are underway at organizations and cause the organization to backtrack. However, changes are unlikely for this reason and because the final stage of meaningful use is nearing a deadline to close the commentary period on May 29, 2015 (Centers for Medicare and Medicaid Services, 2015). This signals a close solidification of the final stage which limits the time for advocacy and for time for changes to occur.

Emerging technologies introduce the possibility that the solution could be simplified by new database management software such as Hadoop, or coding abilities like structured narrative. The planned solution can lose effectiveness by the time of its implementation due to such solutions if they further develop to become more effective as they already appear economical.

Hospital organizations that are experiencing struggles with the use of their EHR or databases will benefit most from this study. In particular, the organizations with disparate applications and unestablished expectations regarding further incorporation of data will gain insights that benefits systems and database planning. It is intended that an organization does not simply take the information contained herein and attempt to apply it, but instead to use this information to form a customized and effective approach that is unique to the organization’s own challenges. Each organization has differing levels of each challenge associated with each barrier and could have separate barriers that are not yet discussed in the scope of this paper, such as acquisitions, mergers, and legal barriers. Organizations must assess their environment and apply this solution where possible to drive a successful outcome.

Prior to pursuing any solution, it is highly advised that organizations determine if there is a need. Is effective use of the EHR lacking? Do separate data sets exist that can improve decision making? Are there plans to expand applications? These are important questions to address at the outset of a decision to develop a comprehensive data set.

Because structured narrative is experimental and has yet to be applied to commercial or production versions of EHRs, further research is recommended. This solution holds the potential to greatly reduce problems that surround a lack of structured data, which allows for the best uses of structured and unstructured data to be leveraged. The application of structured narrative is useful outside of EHRs. Large hospital organizations should immediately set development time aside for software developers to test this experimental approach.

The solution should also press action from healthcare organizations and encourage organizations to voice their need for extensions to incentive payments for meaningful use if they are encountering difficulty reaching the established criteria. Actively voicing concerns will spur important dialogue between healthcare providers and associations that influence policy such as the Centers for Medicare & Medicaid Services (CMS) and lead to more effective policies.

Research supports that expanding the data set utilized by EHRs is achievable, but complicated barriers exist that often prevent organizations from incorporating further data. Each barrier must be approached in a way that is independent, but contributes to an overall solution that addresses all barriers. Analysis has revealed that the rapid implementation of EHRs has contributed to a lack of standards, ineffective adoption practices, and short sighted planning in an attempt to avoid penalties and receive financial incentives. As a result, best practices have been uncovered that are cost effective and will aid healthcare organizations in determining the priorities of projects, barriers in planning, and properly adopting EHRs to improve the use of data and increase the value of further data integration. Finally, whether healthcare organizations plan to pursue further data incorporation at this time or later, research must be conducted to proactively address future constraints of assimilating disparate data sets so that they can be leveraged for improved decision making. In the end, all of society has a stake in how this data is used, and to improve our own healthcare, support must be given to organizations looking to improve how they use this data to everyone’s benefit.

Bahga, Arshdeep and Vijay K. Madisetti. 2013. A Cloud-based Approach for Interoperable Electronic Health Records (EHRs). Journal of Biomedical and Health Informatics. Sept. 894-906. Accessed April 12, 2015. http://0-ieeexplore.ieee.org.bianca.penlib.du.edu/xpls/icp.jsp?arnumber=6497443

Beeler, Patrick Emanuel. 2014. Clinical Decision Support Systems. Swiss Medical Weekly no. 144: w14073. Accessed April 28, 2015. http://jc3th3db7e.search.serialssolutions.com/?ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info:sid/summon.serialssolutions.com&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Clinical+decision+support+systems&rft.jtitle=Swiss+medical+weekly&rft.au=Beeler%2C+Patrick+Emanuel&rft.au=Bates%2C+David+Westfall&rft.au=Hug%2C+Balthasar+Luzius&rft.date=2014&rft.eissn=1424-3997&rft.volume=144&rft.spage=w14073&rft_id=info:pmid/25668157&rft.externalDocID=25668157&paramdict=en-US

Blumenthal, David. 2014. Giving Patients Control of Their EHR Data. Journal of Internal Medicine. Accessed April 7, 2014. http://0-link.springer.com.bianca.penlib.du.edu/article/10.1007/s11606-014-3071-y/fulltext.html

Bowman, Sue. 2013. Impact of Electronic Health Record Systems on Information Integrity: Quality and Safety Implications. Perspectives in Health Information Management no. 10: 1c. Accessed April 20, 2015. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3797550/

Bricon-Souf, N. 2014. Trends on Integrating Framework of Applications or Data. Yearbook of Medical Informatics no. 9.1: 55-57. Accessed April 15, 2015. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4287092/

Campbell, Robert J. 2004. Database Design: What HIM Professionals Need to Know. Perspectives in Health Information Management no. 1: 6. Accessed April 18, 2015. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2047327/

Center for Disease Control and Prevention. National Center for Health Statistics. 2014. Use and Characteristics of Electronic Health Record Systems Among Office-based Physician Practices: United States, 2001-2013. NCHS Data Brief, no. 143. Accessed April 26, 2015. http://www.cdc.gov/nchs/data/databriefs/db143.pdf

Centers for Medicare and Medicaid Services. 2015. Announcement: Stage 3 Rule of Meaningful Use. CMS.gov. Accessed May 2, 2015. http://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/index.html?redirect=/ehrincentiveprograms

CITO Research. 2014. Putting the Data Lake to Work: A Guide to Best Practices. Hortonworks. Issue 4 (April). Accessed April 26, 2015. http://hortonworks.com/wp-content/uploads/2014/05/TeradataHortonworks_Datalake_White-Paper_20140410.pdf

Commins, John 2014. EHR interoperability remains elusive. The Doctors Office. Accessed April 6, 2014. http://0-search.proquest.com.bianca.penlib.du.edu/docview/1467748375?pq-origsite=summon

Connolly, John J., Hakon Hakonarson, M. Geoffrey Hayes, Abel N. Kho, Peggy L. Peissig, Luke V. Rasmussen, and Justin Starren. 2013. Practical challenges in integrating genomic data into the electronic health record. Genetics in Medicine no.15: 772-778. Accessed April 18, 2015. http://jc3th3db7e.search.serialssolutions.com/?ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info:sid/summon.serialssolutions.com&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Practical+challenges+in+integrating+genomic+data+into+the+electronic+health+record&rft.jtitle=Genetics+in+medicine+%3A+official+journal+of+the+American+College+of+Medical+Genetics&rft.au=Kho%2C+Abel+N&rft.au=Rasmussen%2C+Luke+V&rft.au=Connolly%2C+John+J&rft.au=Peissig%2C+Peggy+L&rft.date=2013-10-01&rft.issn=1098-3600&rft.eissn=1530-0366&rft.volume=15&rft.issue=10&rft.spage=772&rft.epage=778&rft_id=info:doi/10.1038%2Fgim.2013.131&rft_id=info:pmid/24071798&rft.externalDBID=5PM&paramdict=en-US

Declerck, G., and X. Aimé. 2014. Reasons (Not) to Spend a Few Billions More on EHRs: How Human Factors Research Can Help. Yearbook of Medical Informatics no. 9.1: 90–96. Accessed April 1, 2015. http://jc3th3db7e.search.serialssolutions.com/?ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info:sid/summon.serialssolutions.com&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=%E2%80%9CBig+Data%E2%80%9D+and+the+Electronic+Health+Record&rft.jtitle=IMIA+Yearbook&rft.au=M.+K.+Ross&rft.au=W.+Wei&rft.au=L.+Ohno-Machado+L.+Zhou&rft.au=Y.+Lu&rft.date=2014&rft.pub=Schattauer+Verlag+f%C3%BCr+Medizin+und+Naturwissenschaften&rft.issn=0943-4747&rft.spage=97&rft.epage=104&rft.externalDocID=http_www_schattauer_de_de_magazine_uebersicht_zeitschriften_a_z_imia_yearbook_imia_yearbook_2014_issue_special_manuscript_22303_show_html&paramdict=en-US

Dictionary.com. 2015. Relational Database. Dictionary.com Unabridged. Accessed May 1, 2015. http://dictionary.reference.com/browse/relational+database

Gartner. 2014. Gartner Says Beware of the Data Lake Fallacy. Gartner. Accessed April 21, 2015. http://www.gartner.com/newsroom/id/2809117

Giokas, Dennis. 2008. Canada’s Interoperable EHR. Journal of AHIMA no 79.5: 56-57, 60. Accessed April 15, 2015. http://0-search.proquest.com.bianca.penlib.du.edu/docview/212624698?pq-origsite=summon

Gooch, Jamie J. 2008. Integration and Prioritized Data Essential for EHRs. Managed Healthcare Executive no. 18.9: 32-33. Accessed April 13, 2015. http://0-search.proquest.com.bianca.penlib.du.edu/docview/212613826/abstract/678EC0222FC94E2DPQ/1?accountid=14608

Haugen, Heather, and Jeffrey Woodside. 2010. Beyond Implementation: A prescription for Lasting EMR Adoption. Magnusson Skor Publishing.

Healthit.gov. 2015. How to Attain Meaningful Use. Accessed April 29, 2015. http://www.healthit.gov/providers-professionals/how-attain-meaningful-use

Healthit.gov. 2015. How much is this going to cost me? Accessed April 27, 2015. http://www.healthit.gov/providers-professionals/how-attain-meaningful-use

Johnson, Stephen B, et al. 2008. An Electronic Health Record Based on Structured Narrative. Journal of the American Medical Informatics Association no. 1: 54-64. Accessed April 29, 2015. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2274868/

KDnuggests. 2015. Awards and Praise for KDnuggets. KDnuggets. Accessed May 1, 2015. http://www.kdnuggets.com/about/awards.html

Lewis, Nicole. 2012. EHR Innovation Gap Threatens Healthcare Progress. InformationWeek Online. Accessed April 12, 2015. http://0-search.proquest.com.bianca.penlib.du.edu/docview/1021342068?pq-origsite=summon

Mai, David. 2014. Is ETL on Life Support? Data Integration Solutions Review: Buyers Guide and Best Practices. Accessed April 29, 2015. http://solutions-review.com/data-integration/is-etl-on-life-support/

Medical Economics. 2012. EHR Connectivity Best Practices. Medical Economics no. 89.18: 66. Accessed April 14, 2015. http://0-search.proquest.com.bianca.penlib.du.edu/docview/1081342559?pq-origsite=summon

Reich, Lance D. 2013. Cognitive Biases. Scitech Lawyer no. 10.1: 5-7. Accessed May 2, 2015. http://0-search.proquest.com.bianca.penlib.du.edu/docview/1464990210?pq-origsite=summon

Rogoski, Richard R. 2012. Bridging the EHR Divide. Healthcare Informatics no. 29.10: 8-10, 12. Accessed April 15, 2015. http://0-search.proquest.com.bianca.penlib.du.edu/docview/1356985339?pq-origsite=summon

Sanchez, Martin F. and K Verspoor. 2014. Big Data in Medicine is Driving Big Changes. Yearbook of Medical Informatics no. 9.1: 14-20. Accessed April 13, 2015. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4287083/

Sanghavi, Sundeep. 2014. Data Lakes vs Data Warehouses. KDnuggets. Accessed April 24, 2015. http://www.kdnuggets.com/2014/06/data-lakes-vs-data-warehouses.html

Studeny Jana, and Alberto Coustasse. 2014. Personal Health Records: Is Rapid Adoption Hindering Interoperability? Perspectives in Health Information Management. Accessed April 1, 2015. http://0-search.proquest.com.bianca.penlib.du.edu/docview/1614637382?pq-origsite=summon

Terry, Ken. 2015. Mining EHR Data for Quality Improvement. Medical Economics no 92.6: 40-43. Accessed April 15, 2015. http://0-search.proquest.com.bianca.penlib.du.edu/docview/1667981991?pq-origsite=summon

Terry, Ken. 2013. New Query Tool Searches EHR Unstructured Data. InformationWeek Online. Accessed April 15, 2015. http://0-search.proquest.com.bianca.penlib.du.edu/docview/1287882172?pq-origsite=summon

Tripathi, Micky. 2012. EHR Evolution. Journal of AHIMA. Accessed April 5, 2015. http://0-search.proquest.com.bianca.penlib.du.edu/docview/1040421420?pq-origsite=summon

Voltz, Donald M. 2015. Connecting the Disparate: Middleware’s Role in Solving Healthcare’s EHR Interoperability Problems. Journal of AHIMA no. 86.5: 28-33. Accessed May 2, 2015. http://library.ahima.org/xpedio/groups/public/documents/ahima/bok1_050918.hcsp?dDocName=bok1_050918