Clinical Documentation: C A L I FOR N I A H EALTH C ARE EHR Deployment Techniques F OU NDATION Introduction n California Rural Indian Health Board This investigation and analysis of clinical (CRIHB) documentation and structured text is the second 3. Multi-site expansion: Support multi-site in a series of tactically oriented issue briefs arising Issue Brief clinics to expand adoption of their existing from the California Networks for EHR Adoption EHR product and implementation services to (CNEA) initiative. The CNEA program was at least three additional clinic sites. Grantees: begun in 2006 to speed adoption and lower the overall cost of electronic health records (EHR) in n Golden Valley Health Centers (GVHC) California community clinics and health centers n Shasta Community Health Center (SCHC) (CCHCs).1 In August, 2008, eight grantees 4. Hospital-based regional extension: Work representing four models of EHR deployment with local hospitals to extend their existing were funded to advance the adoption of EHRs EHR product and implementation services to in the safety net and to share their experiences. CCHCs in a region or service area. Grantees: Through this health IT collaboration, an array of services is provided to support the adoption of n San Mateo Medical Center (SMMC) EHR and other applications. The CNEA models/ n The Children’s Clinic, Serving Children grantees include: and Their Families (TCC), Long Beach, 1. National network: Build or leverage existing California EHR networks, often national in scope, to For additional information about the CNEA provide for individual or groups of clinics in initiative and the collaborative models of adoption, California. Grantees: please see “Making a Connection: Clinics Collaborate n Open Door Community Health Center on EHR Deployment” from the California (ODCHC) in partnership with Our HealthCare Foundation.2 Community Health Information Network (OCHIN) Clinical Documentation Overview n Next Generation Health Network For most of the history of modern medicine, (NGHN) clinical notes about a patient’s case have been circulated locally within a team of primary 2. Clinic consortia: Work with California care providers. For those doctors, nurses, and clinic consortia to expand their existing EHR assistants who are generally all in the same place, product and implementation services to at and who share a technical vocabulary and a set least three of their members. Grantees: of assumptions, handwritten clinical notes are n Redwood Community Health Coalition generally adequate. There are a few drawbacks: (RCHC) The chart may not be available at the point of A pril 2010 Figure 1. Structured Vocabularies as Subset care; the handwriting may not be legible; it may be of Structured Data difficult to aggregate data about a subgroup of patients; and the information may not be complete (for example, tests and vaccinations may not be included). All Clinical Data (including free text) With electronic health records (EHRs), the information can be accurately disseminated to multiple users, Structured Data including specialists, behavioral care providers, labs and pharmacies, insurers, public health entities, and research organizations. In an important departure from paper records, EHRs also can provide decision support at the Structured Vocabularies point of care and aggregate data across large populations to identify health trends and potential adverse effects from new treatments. To fully exploit these capabilities, EHRs must move beyond simply transferring free-text notes into an electronic format; they must capture clinical information with structured data and vocabularies. See Figure 1. For example, a provider recording a visit with a diabetic This issue brief explores the clinical documentation patient using pre-filled templates, drop-down lists, radio options available to users of most EHR systems, the pros buttons, and other pre-determined choices would be and cons of various structured and free-form methods advancing patient care in several ways: of capturing data, and the impact on efficiency and effectiveness of EHR use. n The data are recorded in a searchable format, in which various diabetes-related terms are Methods for Capturing Data intelligently linked to each other. The provider The capture of clinical data exists along a broad or clinic could then find out how many diabetic continuum from free-form typed clinical notes to various patients have current HbA1C test results, or how levels of structured or coded vocabulary and constrained many need foot exams. Structured vocabulary choices. Constrained choices might take the form of allows users to measure the effectiveness of drop-down lists, check boxes, radio buttons, and other data specific treatments. entry devices. Any software that requires data to be entered n The structured vocabulary can be linked to in this kind of specific way is creating structured data. clinician-written decision trees, which prompt additional actions by the provider. For example, The use of structured vocabulary to record diagnoses, in response to a particular clinical term entered by symptoms, surgical procedures, treatments, and drugs3 the provider, the EHR might prompt a referral to multiplies the benefits of EHRs by capturing clinical data a podiatrist, or it might automatically call up the in words or phrases that have standardized definitions. latest linked lab results. In this way, the EHR can Unlike handwritten, dictated, or typed free text, structured support best practices. vocabulary enables the computerized analysis of data for n The data can be linked to other tables, such as the patient care, research, and evidence-based medicine. ICD9 billing codes. 2  |  California HealthCare Foundation n Data can be aggregated and searched, revealing difficulty with the clinical shorthand (DM2, HTN, MI), patterns and trends on multiple levels: for an and would probably miss the significance of a note that individual provider, for a clinic, for disease contained a misspelling. tracking for public health, and for population health management. Using a standardized vocabulary, as in the second case, Free-form Versus Structured the physician is prompted to select certain terms that are The following examples of a history of present illness mapped to codes and to tables of related clinical terms. (HPI) summarize the same encounter between a provider Automated clinical decision support or reporting systems and a patient. The first is written as a free-form clinical would already know the following elements about this note (whether captured on paper or electronically): patient without needing any further information. Mr. P is a 58 y/o AAM with DM2, HTN, CAD with Disorders: CABG who presents now with atypical chest pain, a n Diabetes Mellitus II benign physical exam except for SBP of 160, negative n Essential Hypertension cardiac enzymes and chest x-ray, and nonspecific ST n Coronary Arteriosclerosis changes on EKG. However, given his multiple cardiac risk factors, MI cannot be excluded. n Acute Myocardial Infarction The second HPI is captured through an EHR using Procedures: structured vocabulary terms (in this case SNOMED). n Coronary Artery Bypass Graft x 4 The codes in parentheses are mapped to the vocabulary Findings: terms so that they appear automatically: n Atypical Chest Pain Mr. P is a 58 y/o AAM with “Diabetes Mellitus II” (CID 44054006), “Essential Hypertension” n Non-specific ST-T Abnormality on (CID 59621000), “Coronary Arteriosclerosis” (CID Electrocardiogram 53741008) with Coronary Artery Bypass Grafts x 4 Because the structured vocabulary understands synonyms, (CID 232772009) who presents now with Atypical other providers caring for this patient could use the Chest Pain (CID 102589003), a benign physical exam phrases “heart attack,” “cardiac infarction,” or “MI” except for SBP of 160, Normal Cardiac Enzymes (CID instead of acute myocardial infarction. This eases the 166662009) and Standard Chest X-Ray Normal (CID challenge of providing automated clinical decision support, 168733007), and Nonspecific ST-T Abnormality on and helps improve the coordination of care. Furthermore, Electrocardiogram (CID 42875005). However, given because SNOMED has a detailed set of relationships his multiple cardiac risk factors, Acute Myocardial behind each term, a report could be run to identify all Infarction (CID 57054005) cannot be excluded. patients with “Heart Disease” (CID 56265001). Entering free text, as in the first example, allows the provider to note clinical or intuitive connections that Challenges of Structured Vocabularies might not be anticipated by structured vocabulary. Free However, there are significant challenges with using text also allows the provider to easily add notes about structured vocabularies. The process of capturing all different conditions in the same patient, in this case findings in SNOMED terms can be time-consuming, diabetes and CABG. However, automated reporting depending on the user interface and the availability of systems and clinical decision support would have shortcuts like “favorites” to ease data entry. In addition, there would need to be additional clarification to help the Clinical Documentation: EHR Deployment Techniques  |  3 system recognize that “Acute Myocardial Infarction” was Coded Terminology and Mapping mentioned as a disorder “that cannot be excluded,” rather Most health care providers are familiar with other examples of coded terminology, such than as one that is present. Without such additional as the International Classification of Diseases context, the information captured by structured (ICD), Current Procedural Terminology (CPT), vocabularies is occasionally quite misleading. and Diagnostic and Statistical Manual of Mental Disorders (DSM). Each standard serves a different purpose. The ICD and CPT are used in the United It should be noted that free-form text and structured States primarily for billing purposes, while DSM is a vocabulary are not mutually exclusive. An EHR that classification of mental illnesses, diagnostic criteria, uses structured vocabulary might also have a text box for and treatment options. entering a free-form section of the clinical note. Further, There are two main standards for structured the level of detail that is appropriate in a clinical note vocabulary in EHRs: the Systematized differs depending on who will be using the EHR and Nomenclature of Medical Clinical Terms (SNOMED what data users will want to search for and aggregate. For CT) and the MEDCIN system. SNOMED is managed by the International Health Terminology example, a drop-down list for headache symptoms for Standards Development Organization (IHTSDO), a system used by a neurologist will be far more detailed and is used by the U.S. Department of Health and than in a system used by an internist. Discussions about Human Services. MEDCIN, created by Medicomp the best technological solution will inevitably lead to Systems, Inc., is the foundation for Medicomp products, such as the EMR systems they discussions about best medical practices and about shared developed. MEDCIN maps its terms to SNOMED uses for clinical data. as well as to CPT, ICD, and others. Each of these systems provides coded medical Balancing Structure and Flexibility terminology that can be mapped to other indexes, While structured vocabulary enables efficient decision codes, or tables. Such mapping makes clinical data support and analysis of patient data, flexibility is also available to parts of the health care system beyond the first circle of caregivers by linking data to other important. It is often useful for the clinician to note structured vocabulary standards. For example, ICD- aspects of a patient’s history or correlations of symptoms 9 codes can be linked to CPT codes for billing and not anticipated by a pre-determined structure, and other administrative tasks. Structured vocabulary to follow his or her own thought process during the can also be mapped to research criteria for examination, uninterrupted by questions or prompts utilization reviews, population health research, and vaccination registries. from a structured EHR. In some cases, free-form clinical notes may be faster than using an EHR, as even the most In structured vocabulary, a word or phrase can be linked to several other terms and to detailed thoughtfully designed template will not be appropriate associated information. For example, Myocardial to every clinical situation. Too little detail constrains Infarction has an ID number that is associated the diagnosis, while too much slows the clinician down. with other terms, such as Cardiac Infarction, Finding the right balance is particularly difficult when the Heart Attack, and Infarction of the Heart. (In a configurable EHR, clinicians can add other terms patient has multiple chronic conditions. and tag them with the appropriate ID number.) On other axes, those terms are linked to detailed The table shows some of the trade-offs to consider when information such as signs and symptoms, developing an EHR system. procedures, drugs, and therapies. This information can be used to provide decision support for the provider at the point of care. 4  |  California HealthCare Foundation Table 1. Documentation Methods Compared Method Structured Templates Description Structured templates are partially filled-in notes created in advance for the most common cases seen by a clinic, department, and/or doctor. Such templates allow clinicians to create a note that not only serves as the medical record, but also stores the data as structured vocabulary, which can drive clinical decision-support functions (such as alerts and/or treatment suggestions based on established care guidelines). Advantages Structured templates ease note-taking by presenting a draft to the clinician, who then modifies portions of the note to reflect the current patient’s condition. Depending on the values and content of data fields, additional forms may pop up (as determined by care guidelines of the health center) to ensure a thorough exam. For example, a symptom of chest pain could prompt questions about exertion, family history of heart disease, and history of smoking. Standardized templates allow for uniformity in data capture and in the standards of care provided. Over time, templates improve speed of note-taking and can drive the billing process directly from the clinical record. Disadvantages Physicians often complain that notes from structured templates are difficult to read as a narrative of the patient’s condition, as they appear more as lists than as prose. A “chart by exception” approach, where all findings are pre- populated as negative unless the provider explicitly checks and documents a positive finding, brings its own problem of seeming to provide data the clinician may not have actually captured. Considerable preparation is required to develop structured templates that present enough structured vocabulary to cover the types of patients and conditions seen in the practice, but not so much that clinicians are hobbled, or that the data cannot be classified. Structured templates can be unwieldy for notes about patients with multiple chronic conditions. Best Suited For Structured templates work well for clinical scenarios with an isolated problem that fits a clear framework with a limited number of easily predictable variations. For example, a template for a patient with a chief complaint of headache can present to the provider a set of signs, symptoms, and physical exam findings that should be checked and listed as either present or absent. Similarly, structured templates work well for acute episodes of care for which there are clear clinical guidelines, for specialty care in which the range of diagnoses and conditions is limited, and for predictable well care checkups or OB cases that follow a regular schedule. Method Radio Buttons, Drop-down Lists, Check Boxes Description These data entry methods provide clinicians with a variety of ways to enter structured data. A drop-down list presents suitable alternatives to clinicians and usually offers menu options based on data entered in an earlier part of the note. Radio buttons help enter mutually exclusive data options such as gender or yes/no answers. Check boxes help in maintaining and tracking quality of care by signaling exams performed or questions asked. Each entry selected is encoded and stored as a discrete data element that can be aggregated or can drive other clinical protocols and decision support. Check boxes, drop-down menus, and radio buttons can also be embedded within sections of templates. Advantages These devices allow for rapid point-and-click selection of structured data elements and are especially useful when, through logic imbedded in the EHR, only the appropriate subset of available data selections is displayed. For example, only medications commonly associated with diabetes would be displayed in a drop-down list for a diabetic patient. Disadvantages Some providers complain of too much “clicking” and about the inconvenience of moving from keyboard to mouse to keyboard when documenting their plan of care or clinical note. As with any structured template, careful thought must be given to the appropriate level of detail for choices in lists, check boxes, etc. In addition, there has been some concern that not enough research has been done to determine the optimal spacing between items in a drop-down list to prevent accidental selection of the wrong data element. Best Suited For Radio buttons and check boxes are best suited for yes/no questions such as those about smoking, drinking, or I.V. drug use. Drop-down lists are appropriate for parts of the clinical note where there are only a few possible descriptions, and for ruling out symptoms (e.g., no upper respiratory symptoms, no G.I. symptoms). Clinical Documentation: EHR Deployment Techniques  |  5 Table 1. Documentation Methods Compared (cont.) Method Manual Data Entry (Typing, Stylus, Free-Text Templates) Description Manual data entry is referred to as free text in EHR parlance and can be accomplished by typing on the keyboard or, in some applications, by handwriting with a stylus. Free-text templates allow providers to type in notes in any format. Advantages If the provider is a fast typist, text can be entered quickly and the note will be expressed in the provider’s preferred style of writing. Free-text templates capture notes that can be retrieved electronically and allow providers to include information that doesn’t have a predetermined place in structured vocabulary. Disadvantages Free-text templates are not coded or linked, so they will not trigger prompts or additional information, and they will not show up in searches of linked terms. Best Suited For Free-text templates work well for capturing additional data or interpretations not covered by structured vocabulary, such as parts of the note where there is a significant degree of variation that does not easily lend itself to a pre- defined template. The History of Present Illness and the Past Medical History sections are examples. Method Voice Dictation, Dictation/Transcription Description Traditional dictation/transcription is streamlined in many modern EHRs by having the provider dictate directly into a microphone attached to a computer. A .wav or audio file is created and embedded at the point in the chart that the dictation took place. The .wav file is then sent to a transcriptionist, typed, and uploaded back into the chart as a free- text note. Voice dictation systems translate the spoken word directly into written text within the EHR. Advantages Dictation offers the most rapid method for capturing a fully detailed narrative with the least effort expended. Disadvantages Like free text, dictated information is not structured vocabulary, although recent advances may lead to the ability to capture certain dictated phrases as encoded data. Current voice dictation systems must be “trained” to the provider’s voice, accent, common phrases, and medical specialty terminology. Best Suited For Like free text or manual typing, dictation works well for parts of the note where there is a significant degree of varia- tion that does not easily lend itself to a pre-defined template. Experiences from CNEA Grantees said. RCHC hopes to create specialized audit The following examples from four CNEA grantees reports to identify when this happens and to illustrate several of the points discussed above, including provide support that will encourage providers the tension between free clinical expression and to use the templates. Emphasizing the need to structured data capture. They also demonstrate that the keep all consumers of clinical data in mind, Dr. transition from paper records to electronic standardized Moore acknowledged that templates are hard to documentation is a process that evolves over time and build. “Doctors tend to build templates according requires constant attention. to their personal preferences and to meet their n Dr. Robert Moore, medical director of RCHC and specific needs and vision,” he said. However, he Clinic Ole, noted that the process of configuring added, this can sometimes derail the effort to keep the system never ends; staff is always evaluating more strategic data collection goals at the forefront templates and building new ones. Dr. Moore of the documentation process. anticipates hiring a mid-level provider whose n John Giambruno of Planned Parenthood Mar primary responsibility will be to manage the use Monte, a member of the NGHN, described an and configuration of documentation templates. enlightening moment during the development “It’s easy for providers to subvert the system by of a standard configuration for their EHR: “One just entering things in free text rather than using of the very first ‘Ah ha’ moments was when pre-defined sets of orders or structured text,” he we took the pregnancy testing and counseling 6  |  California HealthCare Foundation visit, the most basic visit we have, took all the care. A thoughtful design process is crucial to developing affiliates’ forms, and found that there was only 30 an EHR that meets the needs of disparate groups without percent commonality. There’s no reason that an adding too many new tasks to an individual’s workflow. organization like ours with a standardized set of That design process needs to result in agreements about protocols needed such a customized approach.” standards of care; it will be most successful if it includes Seeing that a common template would make it representatives from all the groups who will be using the easier to implement the EHR and analyze the data system—those entering clinical notes and those retrieving, led them to secure funding for developing a set of classifying, and linking to the data contained in the notes. standard templates. n Shasta Community Health Center took an evolutionary approach to standardizing templates and clinical documentation. CIO Charles Kitzman Author said, “Without standardization, you don’t get true SA Kushinka, M.B.A., Full Circle Projects, Inc. quality. Constant variation will never give you a picture about what really works.” Initially when they went live with their EHR the focus was A c k n ow l e d g m e n t simply on getting staff comfortable with the new Special thanks to Madeleine Kahn, Ph.D., and Christopher system. Standardization and structured data entry D. Johnson, M.D., M.P.H., for their invaluable assistance. was the next step. Kitzman said they were surprised at the many factors that cause “exceptions to the About the F o u n d at i o n rule,” such as homeless patients and patients with The California HealthCare Foundation is an independent multiple insurance carriers–which require capturing philanthropy committed to improving the way health care unique data elements. The solution for Shasta was is delivered and financed in California. By promoting listening and testing refinements and revisions over innovations in care and broader access to information, our multiple cycles. goal is to ensure that all Californians can get the care they n Several grantees reported convening a clinical need, when they need it, at a price they can afford. For more standards committee or clinical steering information visit www.chcf.org. committee to define and vet clinical templates and documentation standards. Golden Valley Health Endnotes Centers, Shasta, San Mateo Medical Center, 1. The Blue Shield of California Foundation (BSCF), the and NGHN all adopted this approach. While California HealthCare Foundation (CHCF), and the time-consuming initially, it proved to be essential in Community Clinics Initiative (CCI), a joint project of gaining consensus and buy-in for the new system. the Tides Foundation and The California Endowment, are funding this project. Conclusion 2. www.chcf.org/topics/view.cfm?itemid=134138. A smoothly running EHR offers many benefits, but 3. Structured vocabulary is sometimes called coded comes with some costs: Productivity drops during the terminology or controlled vocabulary. implementation period, and the system is only as good as the information entered into it. Structured vocabulary must be a balance between pre-determined data and choices, and the flexibility that is a key to good health Clinical Documentation: EHR Deployment Techniques  |  7