Structure of the tree
As part of an EMR system, we constructed a SDE tree for pediatric patient history and physical examination data (figure 2). The top of the tree in this EMR system starts with history and physical examination. History divides in branches with the specified concepts past medical history (including immunization status), family history, allergies, social history, current medications and the current chief complaint. The chief complaint exists of a general tract and 14 specific history tracts (the respiratory-, circulatory-, gastrointestinal-, and urogenital tract, ear-nose-throat, skin, organs of sense, the nervous-, endocrine-, locomotor- and hematological system, feeding history, prenatal and delivery history, growth and developmental history). Each of these, in total 20, history concepts splits again in 5 to 25 sub branches. For instance, the gastrointestinal tract subdivides in 10 main concepts: general feeding pattern (including undernutrition, overnutrition, and appetite, which is linked to feeding history from the list of specific history tracts), intolerance- or allergy for food, swallowing difficulties, vomiting, nausea, general defecation pattern (including diarrhea, constipation, mucous stools, painful defecation and bloody stools), abdominal pain (acute and chronic), pyrosis, flatulence and fecal incontinence. Sub branches are in the end described by 4 to 15 general attributes, like duration, severity, timing and setting of complaint, influencing factors of complaint and associated manifestations.
Physical examination splits in 11 branches (general survey and vital signs, head and neck, thorax, abdomen, skin, lymph nodes, limbs, genitalia, anus and rectum, nervous system, and spine), of which each forks as well. In total, the thesaurus consists of about 1800 items, used in 8648 nodes in the tree with a maximum depth of 9 levels. Patient history contained 6312 nodes, and physical examination 2336.
Data entering
By selecting a concept (e.g. acute abdominal pain) in the tree, displayed on the left of the screen, a data entry form is displayed on the right with the descriptors (e.g. localization, timing, course) of the selected concept as options for data entry (figure 2). Data entry is accomplished primarily with the mouse: the user selects from pull-down boxes and pick-lists. Concepts can be marked as "present", "absent" or "unknown", and no default was set. Free text annotations can be added to any finding or description. Values, e.g. body temperature, have to be entered using the keyboard. If an implausible value is entered for a particular item, e.g. 378 degrees Celsius for body temperature, a warning appears and the item will be presented in red (figure 3). Clinicians are not forced by the system to change the data.
A search option is available to locate a specific item in the tree. Entering a search term results in a number of hits, including synonyms. Each hit is displayed with its pathway, showing its position in the tree.
To facilitate data entry, we defined several (226) "shortcuts" (reference from one node to another), by which searching and scrolling up and down the tree, as well as repeatedly recording of data in multiple places, are avoided. For example, fever is described in the general history tract. As fever is an important associated manifestation of many symptoms, it has to be addressed in several history tracts as well (e.g. in the history taking of coughing, which is described in the respiratory tract, or in the history taking of vomiting, which is described in the gastrointestinal tract). When the clinician selects, for example, the "shortcut" 'fever' in the respiratory tract, the program directly jumps to the descriptors of fever in the general history tract, where the data on fever will be stored. After completing data entry on fever in the general tract history, it is possible to return to the previous position in the tree (e.g. coughing in the respiratory tract) immediately.
At each moment the clinician is free to choose the starting point and endpoint (degree of detail) of data entry. Moreover, the user can, without affecting the underlying data representation, define an unlimited number of custom entry forms that contain a selection of nodes in the tree for a specific medical problem or disease entity. On such a custom form, medical concepts from different positions in the tree can be combined. Figure 4 shows a custom form for acute abdominal pain that allows the physician to record the signs and symptoms, including localization and timing of the complaint, and physical examination of the abdomen.
The program provides the possibility to export entered data to a text editor or to Microsoft Word. The exported data will comprise basic administrative patient data, the date of data collection, the name of the clinician who recorded the data and the collected data from history taking and physical examination in outline (indented) display. The text can be edited and completed by for instance a description of the interpretation of the presented clinical problem, a diagnosis or a treatment strategy. The resulting report can be used as a letter to the general practitioner.