Canadian Emergency Department Information System (CEDIS) Presenting Complaint List (Version 1.0)
Eric Grafstein, MD;* Bernard Unger, MD;† Michael Bullard, MD;‡ Grant Innes, MD;* for the Canadian Emergency Department Information System (CEDIS) Working Group§
*St. Paul's Hospital, Vancouver, and University of British Columbia, Vancouver, BC, †Sir Mortimer B. Davis-Jewish General Hospital, McGill University, Montreal, Que.‡The University of Alberta Hospital, University of Alberta, Edmonton, Alta.
§For a list of CEDIS Working Group members, please see end of article.
Pressures related to emergency department (ED) overcrowding and cost containment make it increasingly important to characterize the patients we serve and analyze the work we do; but until recently, few EDs have been able to track or describe their case mix, care processes, workloads, utilization, efficiency or patient outcomes. Regional health authorities across Canada have identified electronic data collection as a priority.
Many EDs are developing information systems, but without coordination they are likely to establish different datasets and conflicting data definitions.1,2 Resulting variations in the way that ED data are defined and captured will limit their future utility.3 Recognizing this, the Canadian Association of Emergency Physicians (CAEP), the National Emergency Nurses Affiliation (NENA), and l'Association des médecins d'urgence du Québec (AMUQ) launched the Canadian Emergency Department Information System (CEDIS) initiative -- a program to develop and implement common national ED information gathering systems. In 2001, the CEDIS group published a common national ED dataset2 that EDs and hospital administrators can use as a template for future information gathering.
A system that allows EDs to classify patients and define comparable case-mix groupings will help EDs describe their patient populations, workloads, staffing and resource needs, and enable comparison across sites and regions. Emergency department case-mix groups will be based in part on the Canadian Emergency Department Triage and Acuity Scale (CTAS), which has been prospectively validated and adopted by most Canadian EDs.4,5 However, CTAS defines only acuity; therefore the CEDIS Working Group proposed that ED case-mix groups should be based on both presenting complaint and CTAS triage level. Currently, most Canadian EDs rely on free-text capture of presenting complaints, which precludes categorization, analysis or comparison between sites. Our objective was to develop a standardized presenting complaint list for Canadian EDs that can form the basis for future ED case-mix groups. This would facilitate clinical quality improvement, research and benchmarking at a local, regional and national level.
The CEDIS Working Group is made up of emergency physicians, nurses, administrators and researchers who are active in the field of ED informatics and data management. This group, sanctioned by CAEP, NENA and AMUQ, includes pediatric and adult clinicians, as well as representatives from large and small hospitals from all regions of the country. The formation of the CEDIS Working Group is described in a previous article.2 At a series of meetings in 2001 and 2002, the working group agreed that its first and second priorities were to define standard ED data elements and to compile a common presenting complaint list for Canadian EDs.
At the time of the CEDIS collaboration, 3 affiliated EDs (St. Paul's Hospital [SPH], Vancouver, BC, University of Alberta Hospital, Edmonton, Alta., and Sir Mortimer B. Davis-Jewish General Hospital, Montreal, Que.) were already using electronic presenting complaint lists developed by local emergency physicians, nurse educators, clinicians, health records technicians and information technology personnel. The SPH list was compiled by cataloguing all patient presenting complaints over a 1-year period and then supplementing these with key elements from CTAS,4,5 Emergency Triage (the Manchester Group),6 the Reason for Visit Classification (www.cdc.gov/nchs/data/ahcd/rvc97.pdf; accessed 2002 Nov 22) and the DEEDS (Data Elements for Emergency Department Systems) dataset.7 To ensure that no important presenting complaints were missed, the list was updated after an additional year of triage complaint data collection. Similar methodology was used to develop the 72-item University of Alberta Hospital and the 181-item Sir Mortimer B. Davis-Jewish General Hospital lists. These lists, which have been in clinical use for 4, 1 and 8 years respectively, served as the foundation for the final CEDIS list.
After completion of the adult list, pediatric emergency physicians and nurses were consulted, and the Calgary Children's Hospital ED submitted their list of presenting complaint codes. These were compared to the draft CEDIS list, and several missing pediatric codes were added. One of the authors (E.G.) assembled the final common set, which consists of several major categories with presenting complaints mapping to each.
Throughout the process, care was taken to include only symptoms and presenting complaints -- not diagnoses. For example, a chief complaint of "asthma," would map to "shortness of breath." The collated list was circulated to CEDIS Working Group members for feedback, and adopted at a final meeting in April 2002. A health records technician then attached International Classification of Diseases (ICD)-10 codes8 to the CEDIS presenting complaint list to facilitate accurate data entry. To assess internal validity, one of the authors (E.G.) reviewed actual patient presenting complaints from the SPH site for the year 2002 to determine what proportion would be captured within the CEDIS Presenting Complaint List (Version 1.0).
Table 1 summarizes Version 1.0 of the CEDIS Presenting Complaint List, which incorporates 18 major categories and 161 presenting complaints with their corresponding ICD-10 codes. Two codes, denoted by asterisks, are not actual ICD-10 codes. One is a code created for "minor complaints not otherwise specified" and the other is a code for "traumatic cardiac arrest." We found it difficult to use the ICD system to create an emergency presenting complaint list. Many common ED presenting complaints do not have corresponding ICD-10 codes; therefore, in some cases we assigned the ICD-10 code that most closely approximated the CEDIS presenting complaint. The advantage of using ICD-10 numeric codes linked to the presenting complaint list is that it allows comparisons with other sites or organizations using ICD-10 data and enhances the ability to aggregate information for regional comparisons.
A review of 2002 SPH coding data showed that 98.8% of patient presenting complaints were successfully coded using the proposed CEDIS classification scheme. Retrospective assessment of the complaints that were not successfully coded (e.g., those coded as "minor complaint not otherwise specified") suggested that most, if not all, could have been "fitted" into the CEDIS model. The reasons for lack of accurate coding in these cases included unfamiliarity with the codes by newer triage nurses, the heavy demands of triage at busy times of the day, and patients who had multiple complaints.
Aging populations, acute care cutbacks, hospital closures, overcrowding and the need to improve efficiency have increased the demand for clinical, research and administrative data.9 Several important datasets have been created, including the Canadian Institute for Health Information (CIHI) National Ambulatory Care Registry System (NACRS),10 DEEDS (USA, 1997)6 and the Victorian Emergency Minimal Dataset (VEMD, Australia, 1998).11 These datasets include a dedicated field for ED presenting complaint but do not list a specific set of chief complaints.
In 1979, the US Department of Health, Education, and Welfare published the Reason for Visit Classification, which evolved from earlier classification schemes and contained over 400 specific codes. Unfortunately, this list was created for ambulatory care and its complaints are most relevant to family practice. ICD (the International Classification of Diseases group) and SNOMED® (Systematized Nomenclature of Medicine; www.snomed.org) have also proposed vehicles for presenting complaint use, but these lack a structure suitable for ED use. NACRS,10 the CIHI's dataset for ambulatory care, has been widely implemented in Ontario and some other regions of Canada, but it also focuses on ambulatory care settings and does not provide a presenting complaint field.
In 2001, Aronsky and colleagues12 published what is, to date, the most relevant system for emergency medicine, a coded chief complaint list of 54 items with 3 supplementary free-text fields. These add granularity when necessary and provide information regarding complaints not on the core list of 54 items. Using this system, Aronsky and colleagues were able to reduce the proportion of free-text presenting complaints from 23% to 1%. Although this complaint list allows for grouping of most patients, the rather broad complaint codes offer limited clinical information -- hence the addition of descriptive free-text fields.
Form follows function
Clinicians, administrators and data collectors have different needs, and what a system will be used for determines how it should be structured. When developing a presenting complaint system, it is important to decide whether complaint codes will be used primarily to assist clinicians, to populate a database, to provide mechanism of injury data or to yield diagnostic information.
Clumping and splitting
Increasing the number of unique presenting complaints increases both the specificity and complexity of the system; decreasing the number of complaints enhances reliability and simplicity. If there are too many codes, data analysis becomes increasingly difficult, but if there are too few codes, then not enough information is collected. "Clumping" refers to the notion that some complaints will always be grouped together for analysis and that they should, therefore, be captured as one entity. For example, some patients complain of "headache" but others complain of "migraine." Because triage nurses cannot diagnose migraine, because there is substantial diagnostic crossover between groups, and because the ED treatment is similar, these descriptors should ideally be clumped under the single presenting complaint of "headache."
Consider a patient who falls, sustaining a wrist injury. A "splitter" would code this as "wrist injury" because it gives care providers specific anatomic information. However, if the triage nurse is unsure whether the injury is to the wrist, forearm or hand, and if the injury is misclassified at this stage, the patient will be lost from the appropriate electronic case-mix group. Consequently, to optimize data collection, it might be more appropriate to code the event as "extremity injury." This "clumping" approach makes triage errors unlikely and ensures the patient will not be lost in the database, but provides less clinical information. The appropriate degree of clumping and splitting depends on user needs and the ability to split reliably. To illustrate, dermatologists might design a system with distinct categories for contact dermatitis, seborrheic dermatitis, nummular dermatitis, atopic dermatitis and neurodermatitis; emergency physicians, based on need, simplicity and diagnostic capability, might clump these as "dermatitis." Presenting complaint codes must meet the needs of both clinicians and data managers; therefore, compromises are often required.
Many EDs capture mechanism of injury at the triage desk, and this is for good reason: spontaneous abdominal pain has a different connotation than abdominal pain occurring after a kick from a horse. Unfortunately, adding "traumatic" and "non-traumatic" modifiers for all presenting complaints multiplies the size and complexity of the complaint coding system. Further, some events such as falls, motor vehicle accidents and gunshots tend to involve many body systems and are difficult to succinctly define. The availability of an E (injury) code field in ICD-9 or Sections V to Y in ICD-10 provides more information for clinicians and allows tracking of injury mechanisms without unduly increasing the complexity of the presenting complaint system. Other "additional fields" (e.g. specifying body part or the side affected) may be valuable and add flexibility to the information capture process. These fields may be based on free text or pick-lists.
Diagnosis vs. presenting complaint
It is common to confuse presenting complaint with diagnosis. For example, the CTAS includes a sentinel presenting complaint for "rule-out appendicitis." Clearly, many patients in this category do not have appendicitis and carving them out of the "abdominal pain" case-mix group will hamper subsequent data analysis. Diagnosis is important, but it should be captured separately from presenting complaint.
Structured vs. free text
Free-text presenting complaint systems preclude reliable patient classification, identification and analysis. For example, finding patients with myocardial infarction in a free-text system requires searching for complaints such as chest pain, rule out MI, query MI, ?heart attack, ?cardiac pain and countless others. Assigning all patients with these symptom descriptors a (standard) presenting complaint of chest pain establishes a syndrome-based cohort, or case-mix group, that includes most patients with acute myocardial infarction. In addition, standardized presenting complaints enable the establishment of searchable databases for research and administrative purposes.
Taking this approach means that triage nurses will have to "translate" an infinite range of actual patient descriptors into a limited number of standard complaints. Inevitably, some information will be lost in the process (e.g., "I have gout" becomes "extremity pain"). This is necessary to allow meaningful information capture and subsequent data analysis, but it may initially cause discomfort among nurses who are specifically trained to be scribes rather than translators of patient information. In the new proposed system, it is still possible to record the patient's own words in the ED chart, but this should be done in a separate triage field or in the body of the nurses' or physicians' notes -- not in the standard presenting complaint field.
Defining ED case-mix groupings
Case-mix grouping allows meaningful comparison of morbidity, mortality, complication rates and utilization (e.g., admission rate, length of stay) between physicians, hospitals and regions. Case-mix groups are typically based on diagnosis, and an Australian model proposes that emergency medicine "urgency-related groups" should be based on diagnosis, triage level and disposition.13 But diagnosis may not be the optimal determinant for all ED case-mix groups. Emergency patients generally present with symptoms -- not diagnoses. Most ED processes occur before diagnosis and in many cases a definitive diagnosis is never made (e.g., abdominal pain NYD). When a definitive diagnosis is made, it is not always confirmed during the ED visit. This means case-mix groups based on diagnosis will exclude a substantial proportion of ED patients and ED work.
Emergency case-mix groups based on standardized presenting complaint and CTAS acuity level would yield well defined groups that better characterize the patients, processes, staffing and infrastructure needs of EDs. Several key case-mix groups, representing medical, surgical, psychiatric and trauma related groups could be identified for ED benchmarking purposes. These might include Level I major blunt trauma, Level II chest pain, Level III abdominal pain and Level IV upper extremity injury.
Other emergency case-mix groups may be based on diagnosis instead. This would be used when more specific patient groups are being considered. An example of a diagnosis-based case-mix group might include Level III pneumonias that are discharged.
In the proposed CEDIS system, most pediatric codes are obtained from the adult portion of the presenting complaint list, but several additional pediatric-specific codes have been added. This approach enables adult, pediatric and mixed departments to use a single system. To analyze the pediatric or adult case mix in any given ED, one merely has to sort presenting complaint by patient age.
Limitations and future work
The CEDIS presenting complaint system has great potential, but much work lies ahead. Before this system can be used to define ED case-mix groups, it is important to confirm adequate interobserver classification reliability, and these reliability data are only now being gathered. To enhance reliability, it will be valuable to link as many CEDIS presenting complaints as possible to specific CTAS triage levels, or at least to identify default triage levels that can be overridden based on other modifying patient characteristics. The standardization of these linkages, even for only sentinel presenting complaints would allow across-site comparison of patterns of practice in emergency medicine. Based on feedback from future users, the CEDIS presenting complaint list will require modification and updating. It is important that it become a "living" process.
Finally, the CEDIS presenting complaint list cannot be considered a finished product. It must be a dynamic document that evolves, based on future evaluations of reliability, validity and utility. It must grow if new presenting complaints become important and shrink when "old" codes fall out of use. Monitoring the use of individual codes will allow appropriate updating.
The CEDIS Presenting Complaint List (Version 1.0) represents a compromise between data collection needs and clinical needs of the treating physician. It can be implemented in most EDs without the need for major system upgrades or data download/storage revisions, providing that basic electronic data collection mechanisms are in place. In centres where CEDIS complaints are suboptimal, free-text fields can be added, if necessary, to supplement these presenting complaints.
- Weiss HB, Dill SM, Forjuoh SN, Garrison HG, Cohen JH. Injury surveillance: a statewide survey of emergency department data collection practices. Ann Emerg Med 1996;28:635-40.
- Innes G, Murray M, Grafstein E, for the Canadian Emergency Department Information System (CEDIS) Working Group. A consensus-based process to define standard national data elements for a Canadian emergency department information system. CJEM 2001;3(4):277-84.
- Cordell W, Overhage JM, Waekerle, JF. Strategies for improving information management in emergency medicine to meet clinical, research, and administrative needs. Ann Emerg Med 1998;31:172-8.
- Beveridge R, Clarke B, Janes L, Savage N, Thompson J, Dodd G, et al. Canadian Emergency Department Triage and Acuity Scale: implementation guidelines. CJEM 1999;1(3 suppl).
- Beveridge R. CAEP issues. The Canadian Triage and Acuity Scale: a new and critical element in health care reform. Canadian Association of Emergency Physicians. J Emerg Med 1998;16(3):507-11.
- Emergency Triage. Manchester Triage Group. Mackway-Jones K, editor. London: BMJ Publishing Group. 1996.
- Pollock DA, Adams DL, Bernardo LM, Bradley V, Brandt MD, Davis TE, et al. Data elements for emergency department systems, release 1.0 (DEEDS): a summary report. DEEDS Writing Committee. Ann Emerg Med 1998;31(2):264-73.
- International Statistical Classification of Diseases and Related Health Problems, 10th rev. Geneva: World Health Organization; 1992 (vol 1: tabular list, vol 2: instruction manual, vol 3: alphabetical index).
- Teich JM. Information systems support for emergency medicine. Ann Emerg Med 1998;31:304-8.
- Canadian Institute for Health Information. National Ambulatory Care Reporting System (NACRS) Implementation. Available: http://secure.cihi.ca/cihiweb/en/downloads/services_nacrs_e_elements.pdf (accessed 2002 Nov 1).
- Victorian Emergency Minimum Dataset (VEMD) User Manual, Version 6.0, 2001/2002. Victoria (Australia): Acute Health Division, Department of Human Services; July 2001. Available: hdds.health.vic.gov.au/vemd/index.htm (accessed 2002 Nov 26).
- Aronsky D, Kendall D, Merkley K, James BC, Haug PJ. A comprehensive set of coded chief complaints for the emergency department. Acad Emerg Med 2001;8(10):980-9.
- Cleary MI, Ashby RH, Jelinek GA, Lagaida R. The future of casemix in emergency medicine and ambulatory care. Med J Aust 1994;161 (suppl):S30-3.
CEDIS Working Group: François Bélanger (Calgary, Alta.), Jerry Bell (Regina, Sask.), Gary Bota (Sudbury, Ont.), Michael Bullard (Edmonton, Alta.), Alan Campbell (Mississauga, Ont.), Dan Cass (Toronto, Ont.), Eric Grafstein (Vancouver, BC), Brian Holroyd (Edmonton, Alta.), Michael Howlett (Truro, NS), Grant Innes (Vancouver, BC), Michael Murray, Chairman (Barrie, Ont.), Julien Poitras (Levis, Que.), Brian Rowe (Edmonton, Alta.), Bob Sweetland (Winnipeg, Man.), Bernard Unger (Montreal, Que.), Pat Walsh (Grand Falls/Windsor, Nfld.).