My name is Betsy Haines. I am a psychiatrist and an electronic medical record (EMR) implementation veteran. I am starting "Getting to Go Live" to help other people involved in implementing an EMR lead their organizations through a successful launch. The observations, lessons learned, best practices and thoughts I will share are based in my broad range of experience. 

I spent my 30+ year career on the clinical faculty of the University of Medicine and Dentistry of NJ. For 25 years, I did direct patient care with people with serious and persistent mental illness. My absolutely best days were when I saw patients in the railroad station and did initial contacts in McDonalds. I also supervised professionals of every disciplinary stripe and held a number of administrative positions.

Then for 6 years, I morphed into the project manager for an electronic health record (EHR) implementation. (BTW Aficionados hold to the distinctions between an EHR and an EMR, electronic medical record. We will use the terms interchangeably.)  The implementation was complex, arduous, challenging – your average EHR implementation. Happily, it was also successful.

I recently did a presentation on electronic medical record content design for a customer group. As part of the presentation I gave some tips on dictionary design. The  participants found these useful, so I decided to share them in a blog.

#1 Don’t go beyond your degree of certainty. 

When you are capturing data based on direct observation, “Yes/No” formats are appropriate.

 

But when you are capturing information based on what the client reports, then you cannot use a” Yes/No” format since “No” is unconfirmed. Use “Denied” instead.

 

#2  Provide for clinical judgment.

 When the client denies something crucial, include a choice that supports a different clinical impression. Use “Suspected.”

When the client says “Yes” to something, provide for clinician doubts.

#3 Think twice before giving an opt-out on required items.

Our clinical design team members felt strongly that access to firearms was essential information for the assessment of intentional and accidental risk. We all felt that to include a “N/A” choice would invite clinicians to dodge the question. So despite protests from the geriatric representatives, we made it required without an opt-out choice.

In other situations, however, the balance between importance and exceptions tips towards the latter and opt-outs simply make sense.

For example, a history of abuse is crucial to treatment, but such information may not be acknowledged or even be appropriate for initial inquiry, so an “Unknown” choice makes sense.

In an earlier blog I described our design of an integrated assessment. As part of the design, this Functional Assessment was to be used by the child clinicians. In order for the integrated approach to work, we had to adjust the responses to be appropriate for younger children. Hence the “N/A” choice.

#4 Be sure to include a way-out on required pick lists when none of the selections may apply.

Dictionaries are one of the most useful tools in your design kit. They are vital to capturing structured data in your EMR. They contribute to efficiency and consistency and facilitate robust, aggregated reporting. But use them thoughtfully.

Now it's your turn. Have you got any tips to share with us?

I just read an article on the CIO blog by Rick Cook about “velocity” as an important component of successful IT projects. Velocity is a key concept in agile development. It refers to producing a lot of small deliverables at frequent intervals. Cook describes that velocity makes tracking progress easier and that it's also good psychology since it reinforces a feeling of success and builds team morale. He further writes that IT projects are long and hard to maintain.  You need the frequent small rewards of hitting smaller milestones. 

Cook piqued my curiosity about agile project management, so I dug a bit further into it.  I discovered that the article vastly oversimplifies the discipline. I, for my part, am going to use it loosely to apply to an EMR implementation.  Velocity is a version of creating and maintaining momentum over the course of an implementation, something in which I ardently believe.  The focus on breaking projects into small chunks and delivering pieces of it fast for user feedback keeps the visibility of the implementation high.  Here’s a graphical presentation with traditional waterfall development model in black and agile project implementation in red.

Continuous planning focuses on producing these frequent products for immediate feedback. The development team then reacts to the feedback rather than waiting until the entire application is released. Some examples of chunks in an EMR implementation are specifications for a new assessment, screen shots of new progress notes, or, best of all, a care planner that users can test.

This “agile” process should be designed to involve a range of stakeholders and end-users. It thus also advances my top concern in an EHR implementation – user buy-in. Looks like something worth trying. I would if I were implementing an electronic health record now.

As you have probably gathered from my recent blogs, I am enthusiastic about clinician collaboration in capturing the clinical history in an electronic medical record. In different entries, I’ve covered the thinking behind the idea, “how-to” software procedures and even the emotional response of clinicians. Now I want to look at the crucial question of how is accountability for content maintained? Specifically, how will the system identify the author of the content and the date on which it was filed. The “how-to” entry detailed a procedure, which providers would follow to attach this information to content to be copied forward. The procedure depends on provider action. It is not automatic. Therefore, it would be necessary to monitor compliance with the procedure.

An article in the Jan ’09 Journal of the American Health Information Management Association, suggests these steps:

·       Determine how providers should use the copy functionality

·       Test the procedures

·       Develop a policy that covers the clinician’s obligations to follow the procedures

·       Design the audit plan with schedules and reports

·       Execute the plan

Now here comes one of my "great pronouncements" J about EMR Implementations: Software cannot substitute for monitoring the quality and integrity of content. And certainly no computer can intervene when supervisory follow-up is needed.

In my previous blog, I wrote about how a team of clinicians can collaborate on capturing a clinical history.  The integrated Core Assessment I described replaces separate psychiatric and psycho-social evaluations.  I’ve previously outlined the thinking behind this integrated design.

During our original EMR implementation in 1999, a number of clinicians, especially psychiatrists, expressed unease and resistance about giving up “my” evaluation. I have understood these feelings to represent a combination of professional responsibility and professional narcissism.  I’m going to address the responsibility and accountability topic in my next blog.  For now let’s focus on the question I would ask them, “Whose evaluation is it, yours or the patient’s?”  

This question reliably caused the clinicians to pause, think and give a wry smile.  In presentations to clinical audiences about EMR design, when I would pose this question, a twitter of recognition always passed through the gathering.  The clinicians grasped that the design offered clear benefits of completeness of and ready access to historical information. They appreciated the efficiencies in data entry.  Most acknowledged that the integrated design was the better path, but they felt the loss. The design called for a submersion of disciplinary identity into a team product.  They felt this, gave that smile, took a deep breath and said, “OK.”

That was 10 years ago. Today we are in the time of client-centered care. Maybe the question has a different resonance now. Maybe the “Well, it’s the client’s” is a no-brainer answer.  Is it?

Content design is a crucial component of any EMR implementation. In my last entry, I described a chart design in which assessment information is captured in one form/option, a Core Assessment. The design was created to support multi-disciplinary treatment teams working in acute, brief and long-term settings. In this design, the Core Assessment is the central repository for all the clinical history. It is used by all clinicians at the time of an initial assessment and for updates as historical information emerges in the course of treatment. The Core replaces the separate psychiatric and psychosocial assessments. My earlier blog describes the thinking behind this integrated design.

Sounds OK, but at a practical level, how does this work? I’m going to address the issues in a couple of blogs, as I want to get down to the details, where the devil resides. The first set of questions ask: With multiple authors, who enters data and when? How is the option set-up? How does the historical update work? The solution I’m describing was built in the Netsmart Avatar CWS using the RAD modeling tool. But the planning and setup could apply to other applications with a customization tool set. Also I’m going to give a brief description here. There is a much more detailed scenario with screen shots available in this document.

Here are the issues/questions you will need to address:

·       Who will be entering information into the Core? This is most easily thought about as roles or disciplines?

·       What information are they responsible for? This must be very specific, at the level of individual fields.

·       What fields should be required and for whom?

·       Are there fields which should be electively available for document building?

·       Which fields are History-oriented and should be available for updates? Which are Presentation-oriented and should be protected from updates?

An example solution to these questions:

Then proceed with defining the setup. For the above combination of users, you would:

·       At the very beginning identify the user role or the update function e.g. create a field titled TYPE OF ASSESSMENT. A single response dictionary with these choices: Psychiatrist, Social Worker, Nurse and History Update. Make it required. Note the Case-manager role is not included since their only input is to update history and, depending on policy, any one can enter a History update.

·       Program the logic to support the role-based decisions outlined above.

·       Program the logic to enforce the separation of History-oriented and Presentation-oriented fields i.e. when History Update is selected, all Presentation-oriented fields should be disabled and made inaccessible, while all History-oriented fields should be enabled but not required.

·       HINT Make a spreadsheet matrix of all Type of Assessment items and the individual fields. This is complicated and detailed stuff and needs systematic tracking.

During my years working with seriously and persistently mentally ill patients, I would encounter people who did not fit their given diagnosis or who did not improve despite seemingly adequate treatment or who resisted all efforts to engage them in treatment or who…I could go on. These situations really begged for greater understanding of their histories and courses of treatment. This would mean going through many years of treatment  contained in stacks of paper charts routinely 6-12 inches high.  Sometimes an answer lay in an overall pattern that emerged, but sometimes it lay is a piece of history buried in a progress note. (In my experience, the most missed diagnoses were severe OCD and dissociative symptoms and the most missed pieces of history involved abuse.) Such chart reviews are very time-consuming and, therefore, basically impractical in the modern environment.

The electronic format provides opportunities to mitigate this barrier to good care. At UBHC, as part of our EMR implementation, we designed information flow and data entry procedures that fostered building a person's history and capturing significant nuggets over time.  (This document illustrates the following discussion.) First, we created a Core Assessment which was designed to be the central repository of all historical information on an individual. All clinicians would enter the basic assessment data in the Core Assessment. In other words, we integrated the traditional psychiatric and psycho-social assessments. Physical health and nursing content remained as separate assessments. Next, we conceptually divided the Core content into two types. Presentation-oriented content concerned the immediate information about the patient at the time He/she came for services. Examples are Chief Complaint / Presenting Problem and Current Suicidal Thoughts and Behavior. Such slice-in-time content would not be copied forward or "defaulted" in Avatar lingo. The other type was history-oriented content, such as Personal and Family Histories. This content concerns the patient's life story and maintains integrity over time. History-oriented content would be copied forward indefinitely, including across admissions thus making it available, in one place, to future providers.

Lastly, the historical type of information is revealed in increasing detail as the treatment evolves. Some examples are a history of abuse and the loss of a significant, other-than-parent figure. This is the important information that is typically buried in a progress note and lost to all but the most diligent or lucky of future providers. So we established the procedure that when new history came to light, the clinician should add it into the Core Assessment as a history update. In this way, the historical type content was built as an ongoing collaboration by all providers.

Of course, there are numerous details and other concerns e.g. data entry workflow and accountability which I'll address in future blogs. This seems like more than enough for now. I've tried to present this clearly. Did I succeed? Do you have questions? What do you think?

This is a little off the topic of EMR implementation, but I just saw a website which I thought was so well done that I had to tell you about it.  It is called HMSA Online. It is the brand new telehealth service from HMSA (Hawaii Medical Service Association) the Blue Cross Blue Shield Plan of Hawaii. It integrates

·       24/7 access to a queue of primary care and specialty physicians

·       automated system compatibility check

·       web-based training

·       an EMR

·       a PHR

·       tele-conferencing

·       granular level privacy controls and permissions

·       a Task List

·       a consumer portal

·       e-prescribing

These features and more combine to create a rich-media virtual encounter.

The website is clear and easy to use.  The design of the service from both the patient and physician points of view is elegantly simple and sensible.  It is integrated with the HMSA EMR as well as Microsoft HealthVault. It provides new levels of transparency (the patient sees what the Dr is writing whether note, prescription or suggestion for follow-up care) and of privacy control (patients give access permissions at the granular level of individual lab tests and encounter notes.) 

I could go on and on, but to really appreciate it you have to a look yourself.  To me this looks like a piece of the future of health care. This brochure gives a good description, especially of the Doctor's benefits and procedures.  Please take the time; have a look and let me know what you think. 

I want to piggyback on Chuck Klein’s recent blog in which he makes the case that e-prescribing software is essential for an organization to understand the prescribing patterns of its physicians. There is currently much belief in the land that, given the financial crisis and the new Obama administration, health care reform will actually begin to occur. (I wrote this blog yesterday Jan 7^th. This morning Jan 8^th, I just heard President-Elect Obama state that all Americans will have their health information kept on a computer WITHIN FIVE YEARS!)

Accountability and transparency about costs and results are at the center of proposed fixes. The development of evidence-based treatments is another pillar of health care improvement plans.  Both these initiatives require masses of data. We are not talking about data gathered as part of a study, but data routinely generated in the course of delivering care.  An electronic medical record system is the only hope for capturing such data.

But there is more to it than just an EMR implementation. The content of the EMR has to be designed so that information is in discrete data formats i.e. dictionary based lists and scales. In short, if you don’t click it, you can’t report from it. 

I wrote two previous blogs thinking out loud about typing versus clicking and data type combinations from the clinician’s point of view. But the gathering forces behind EMR implementation are all interested in data. Here’s a blog by Peter Beck that urges clinicians to get over their attachment to narratives, to forget about copy and paste, to stop thinking their efficiency is the top priority. Clearly, tremendous attention is going to be given to EMR implementation. Like it or not, it will all be about creating data.

I just read an article which makes me think that by the time we get our EHR implementations completed, our patients will have leapfrogged us into Healthcare 2.0. 

A Pew Research Center survey found that 80% of adult Internet users in the U.S.—that’s 113 million people—searched for health information in 2006. Increasingly, they are moving beyond passive searches to interactive forums, blogs, and other social media, a trend often referred to as Healthcare 2.0.

Healthcare 2.0 is the use of social media and other technologies to improve communication in healthcare. These platforms may be used to connect patients with patients, ">doctors with other professionals, or patients with doctors. An interesting example is the website Patients Like Me. This site lets people with a limited range of desease, such as Amyotropic Lateral Sclerosis (ALS) do interesting things together. They can by-pass drug companies to report medication side-effects. The ALS group even banded together to run their own clinical trial of Lithium without any external resources. Also researchers can use such websites to recruit subjects for their studies.

There is not yet much content developed in the major mental illnesses, but watch this video (sorry about the pasta ad first) and imagine the potential!

I just read an IDC whitepaper, which advocates thorough training of the IT implementation team.  The discussion is based on a survey of 144 senior IT managers examining success factors of 377 IT projects. To begin with, the data (see graph below) show that End-user software projects have a 78% success rate, which makes them a lower-risk type of implementation. This should be reassuring to people embarking on an EMR implementation. Also note that Business and disaster recovery planning projects are high risk. I mention this in passing because, for all its challenges, disaster recovery planning is a major responsibility of all electronic medical record implementations.

The paper then looks at factors with the most impact on project success. The analysis is shown in the graph below. Keep in mind that the survey respondents were IT managers, which must skew the results towards IT factors. That said, the results indicate that the managers consider the skill of the project team to be most important.

Supporting data show that successful projects spent twice as much time (20 hrs vs 10 hrs) and money (6-10% vs >4% of budget) training their teams than did unsuccessful projects. The authors emphasize that the training must be practical with realistic examples, stating that general skills have little impact on success.

In second place on the list is the dedication of the project team. Executive support and Available resources fall well down the list. These latter placements run contrary to all that I know about change management and again surely reflect the source of the data. However, it is a worthwhile message to focus early on thoroughly training the IT staff responsible for implementing, maintaining and supporting the software.

As mental health clinicians, we place great importance on nurturing a therapeutic alliance with our clients. This involves focusing meaningful attention on the client during the session.  When we were planning our EMR implementation, we worried about how the presence of a computer would impact the relationships with our clients. The answer, of course, is that it depends on the client – and on the clinician.

For 25 years I worked with seriously and persistently mentally ill people. Communicating respect and concern was at the core of the therapeutic effort. Engendering trust was often my most challenging task. I recorded only minimal information during sessions in the paper era and none on the computer. I guess I and, I believed, my patients were just too “pre-computer” to experience interacting with a computer as part of routine flow. In fact, as project manager of the EHR implementation, I persuaded our clinical leadership to commit to allocating time for data entry in addition to the face-to-face interview.

However, my colleagues with clients employed in the modern workplace and who often received health care in managed care settings had a different approach. They found that these clients were accustomed to computers. They even expected to have clinical information captured electronically. So then it comes down to how you do it. Here’s a vignette I came across recently:

This clinician is acutely aware of the hazard of having a machine between the himself and his patient.  "If you spend your entire time looking at the screen, it is going to affect the person’s perception of quality -- and I think it ultimately affects the care you can deliver," he says. He knows that part of his success depends on simple human contact. Consequently, he tries to use the machine as a tool of engagement, not a barrier. When he is interviewing a patient, he periodically stops for a machine-gun burst of typing, during which he repeats out loud what he is writing -- a kind of verbal spell-check. He occasionally turns the monitor to show the person a graph or an image. He ignores the computer during the most important parts of the conversation. "They know they are going to lose some of my attention, but in the end get more information," he says. He types for a few minutes, and then faces the patient again to propose a plan.

This process makes lots of sense to me. What do you think? Do you do something like this? If not, would you try it? Please let me know. Thanks.

The current issue of the Archives of Internal Medicine has an article with some preliminary good news about a positive impact of EHR’s. Based on a survey of EHR use by physicians and on information about paid malpractice claims, a group from Harvard found that physicians with EHR’s had fewer paid malpractice claims.

The data showed that 379 of 1140 (33.2%) of physicians used an EMR. Of physicians with an EHR, a total of 6.1% of physicians had a history of a paid malpractice claim compared with 10.8% of physicians without EHRs (unadjusted odds ratio, 0.54; 95% confidence interval, 0.33-0.86; P = .01). After adjustment for several factors, the difference became not significant. However, the trend was supported by the additional finding that among EHR adopters, 5.7% of physicians identified as "high users" of EHR had paid malpractice claims compared with 12.1% of "low users".

The authors state that “Although the results of this study are inconclusive, physicians with EHRs appear less likely to have paid malpractice claims. Confirmatory studies are needed before these results can have policy implications.”

If and, more likely, when these results are confirmed, another incentive will join the business case for undertaking an electronic health record implementation. Good news.

I am a member of the CCHIT Behavioral Health workgroup. Our task is to develop functionality standards for behavioral health EMR’s. A recurrent theme in our discussions is the lack of content standards in the behavioral health domain. Well, a small step has been taken.

On Nov 13^th the Department of Health and Human Services (HHS) released the first-ever inventory of quality measures that are used for reporting, payment, or quality improvement by its agencies and operating divisions. The Department states its goal is to further collaboration among members of the quality community and to advance the effective use and harmonization of quality of care measures. The inventory is available on the clearinghouse Web site.

The areas covered include alcohol and substance abuse, tobacco use, domestic violence and nutrition and exercise. So sadly there is no guidance on measures for major mental illness. Also, they focus on process rather than outcomes, but it is a beginning. Surely it behooves content designers, especially in the substance abuse area to consider how to integrate these measures into the electronic health records.

Today I want to be less pedantic than I have been recently and have some fun. I've already written about John Kotter in my blog “A Sense of Urgency”. Now I have gone ahead and read his fable about change, Our Iceberg is Melting. It is about a penguin colony in Antarctica. A group of penguins live as they have for many years. Then one curious bird discovers a potentially devastating problem threatening their home. Initially hardly anyone listens. The fable tells about resistance to change and how dramatic change was successfully brought about. The characters are familiar in our own lives and work. You can even decide which character in the drama describes you best.

Kotter believes in the power of stories to communicate important ideas with emotional impact. Read this story. Share it with others on your team. Talk about his Eight Steps and how to accomplish them in your electronic health record implementation. There is even a website with more tips and, of course, materials for sale.

When designing the content of your electronic health record, you have lots to think about. I've just written about data format decisions with some examples.  Databases also permit you to place rule-based controls around Clinician use of options and fields in the EMR. These tools protect data integrity by preventing unauthorized input and protecting content from modification. They help ensure that Clinicians are capturing the same sorts of information and that you are getting the information your organization must have. An effective controls strategy enforces desired content while minimizing the inclusion of inappropriate and unnecessary data. The type and extent of controls also implement your organization's goals and policies.

In other words, they are a very useful set of functions in your design toolkit. Now, what do I mean by Controls?  Here is a list of some with examples:

·    Define access to options and fields as for input or "read only"– Denying input access to finalized forms protects the integrity of the content and of accountability. Denying access to (disabling) a time field that auto-fills with the current time prevents fudging the time vital signs were recorded.

·    Make fields required or optional for filing. Required fields ensure completeness and consistency. Optional fields invite the addition of relevant information in consistent formats.

·   Use event logic e.g., "If this response is selected, then this will happen." If the required item “Pain Present Y/N” is clicked “No,” then the initially disabled fields for Intensity and Location remain disabled, preventing extraneous data. If  “Yes” is selected, then the Intensity and Location fields become enabled and required, ensuring compliance with organizational policies about assessing pain in all patients.

Hopefully I am giving you the idea. This document shows more examples of Controls with Avatar screen shots.

Now that I have just advocated for EMR content customization, let’s return to content design. This is my favorite part of EMR implementation. I wrote earlier on educating the clinician members of your design team about databases. Next they need to understand the different data types available for capturing information and the implications of data type decisions. 

The most important distinction is between free text and dictionary-based data types. Everyone is already familiar with the notion of typing ongoing text. Similarly, most people have encountered check boxes on forms, e.g. medical histories attached to clipboards in doctor waiting rooms. Clinicians are comfortable with creating narratives as part of clinical histories and other documentation. Indeed I have seen several behavioral health EHR’s that consist entirely of textboxes for entering the organizations’ usual narrative content. Such EMR’s are essentially electronic paper, but what a waste! Among the advantages of an EMR are efficiency, standardization and aggregated reporting.  None of these is achieved using narrative textboxes.

But deeper issues lie beneath the type or click choice. Clinicians think of the clinical history as the patient’s story. I heartily agree with this. Effective treatment is grounded in understanding the story of the person’s problems and suffering. At the same time, the story includes numerous individual pieces of information, for example, past substance abuse, the presence of a firearm in the house, sexual abuse by a family member rather than a stranger and a parent who committed suicide. Such salient features of client stories, alone or in combination, can serve to trigger alerts about individual risk. They can help supervisors and administrators identify and monitor the care of clients with a defined set of problems. Aggregated, they can guide managers to unmet service needs.

Likewise, a case summary can be a text formulation and/or lists and measurement scales. The former presses the clinician to integrate an explanatory narrative that justifies an approach to treatment. The latter asks the clinician to abstract specific problems and strengths, to assign a diagnosis and to quantify patient attributes in various domains. The narrative facilitates reflection and understanding. The lists and scales sharpen focus and enable accountability. Aggregated lists and scales can help address important questions including about outcomes, efficacy and best use of resources.

Likewise with progress notes and so on. Now what of the original question – to type or to click? The answer is to do both in artful, disciplined combinations. In my next blog, I’ll give some examples of such combinations, which I hope will be artful and disciplined or at least useful. 

I realize that I jumped right in writing about EMR design. Without thinking, I just assumed that you would choose to develop content for your EMR that was customized by and for your organization. Obviously, you can opt for using an Off the Shelf solution. Let’s back-up a step and look at the pros and cons of content customization.

First the cons: these boil down to time and money. As I wrote earlier, content design is a team effort. It takes many meetings of a diverse staff group to develop a good design. (At UBHC the design workgroup had 29 members who met for a 1/2 day per week for three months.) The staff time costs money. There is no way around that. Also the time involved may extend your implementation schedule. Though this extension is not inevitable if you start the design task early. You can begin even before you select the Electronic Health Record application.

The pros boil down to facilitating staff buy-in and getting a better EHR, one that really reflects the information needs, workflows, clinical culture and policies of your organization. These issues are especially relevant to behavioral health organizations where humanistic values are high and industry standardization is low relative to physical health.

In my experience, leaders of EMR implementations define their greatest concern to be staff buy-in; so why pass-up the primary opportunity to engage clinical staff. I also have seen that those organizations, which began with an Off the Shelf approach, were invariably dissatisfied down the road. They turned to customization and then had to face the more difficult task of retrofitting their new information flows to accommodate what they already had in place. At UBHC the initial design remains sound eight years after the implementation.

So, is customization worth the effort? Clearly, I think it is.

As you already know from a previous blog, I favor defining firm, challenging timelines in an EMR implementation.  These create intensity, momentum and discipline.

In his latest management book, John P. Kotter describes the paramount importance of people within an organization having "A Sense of Urgency." (Check out below a brief presentation of his concepts.) In part of the book, he focuses on the importance of engaging employees emotionally in a change project.  He cautions that a well-devised plan arouses little urgency in an organization. He goes on to explain that human brains are programmed much more for stories than for abstract ideas. He tells of using stories to create an mind/emotion understanding of the risks to the company and the employees should they not move forward.  

In my psychodynamic therapist hat, I know that intellectual insight alone does not bring about change. The mind content needs to be directly linked to an emotional experience for an "Ah Hah!" to happen.


I’m going to revisit my presentations to try for less PowerPoint and more stories.