Structured classification for ED presenting complaints – from free text field-based approach to ICPC-2 ED application

When a patient enters an emergency department (ED), important decisions are made at the very beginning of the visit concerning the necessity for, and the urgency of, medical examinations and care required by the patient. The presenting complaint, other anamnesis available, and a short status assessment determines the urgency of the treatment. Information from the referral notes or ambulance staff is also useful, if available. The presenting complaint is the patient’s reason for the encounter, interpreted and recorded by the triage nurse. Although the importance of triage on patients’ prognoses is recognized [1], there exists only sparse data on the presenting complaints of ED patients.

Over the last 20 years, hospitals have developed different formal triage systems and these systems are in wide use [1]. No golden standard for triage exists, and careful attention should be paid on the studies, and the follow-ups, of the competence and validity of the triage process in each ED. For example, studies on children [2], patients with unspecific complaints [3], and patients with sepsis [4] have shown the difficulties of triage.

To be able to study the impact of triage on the prognoses of different patient groups, structured information of the presenting complaints is needed. The process, from presenting complaint to diagnosis, is the core competence of emergency departments and failures at the beginning of the process - specifically, in triage - often lead to prolonged visits, endangered patient safety, and decreased patient satisfaction. Designing and controlling emergency departments with end diagnosis based information may result in inappropriate processes. 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 [5]. In addition, it allows the development of automated decision support systems to specific patient groups, such as the automated evaluation of the Pneumonia Severity Index [6], or the reminder of stroke assessment form [7]. However, such a classification system should be relatively easy to adopt and implement, in order to be routinely utilized by ED staff.

No international standard for recording presenting complaints exists, and in many EDs, recording is not systematic. Even though the need for a systematic way to record presenting complaints was raised over a decade ago by Aronsky and colleagues [7], studies around the world report vast use of free-text fields and a general lack of a structured way to record presenting complaints (Australia [8, 9]; US [10]; Finland [11]). The Canadian ED Information System (CEDIS) Working Group’s Presenting Complaint List [5, 12] linked to the Canadian Triage and Acuity Scale (CTAS) has achieved good coverage in Canadian EDs. Other large international triage systems (ATS, MTS, ESI) do not provide structured classifications for PCs. Other than CEDIS, there are few other classifications presented in literature. Aronsky et al. [7] aimed at developing a generally applicable set of coded chief complaints for EDs and their study resulted in a list of 54 presenting complaints. Other classifications are presented in relation to single studies [3, 8, 13‐17].

In contrast to fragmentation of presenting complaint classifications in EDs, there are established classifications for ambulatory care. Reason for Visit Classification (RVC) is used by the Centres for Disease Control and Prevention (CDC) and for the annually reported National Hospital Ambulatory Care Survey (NHAMCS) in United States [18, 19]. The International Classification for Primary Care (ICPC-2) is used for both diagnostic classification in family practice and primary care, and classification of ambulatory patients’ Reasons for Encounter. WHO has accepted ICPC-2 within the WHO’s Family of International Classifications (FIC), mainly as a method of encounter classification. ICPC-2 has been used in few studies within emergency services [20] and EDs [21, 22]. However, neither RVC nor the ICPC-2 is ideal for use in EDs. Both systems are designed for primary care office hours and they each include several hundred complaints. (RVC includes 770 different complaints and the shortened version of the ICPC-2 has 687 codes for RFEs.)

The aim of the current study was to investigate whether the ICPC-2 could be adjusted to an ED presenting complaint classification, and thus, achieve an ED classification compatible with other areas of health care using the ICD-10 and the ICPC-2. The main reason for selecting the ICPC-2 was to employ an already internationally established classification as a foundation for the new classification. This study presents a structured classification for adult patients’ ED presenting complaints based on ICPC-2 classification. The authors developed the ED presenting complaint classification by using free text analysis of 40,610 visits in one ED and by collecting presenting complaint data, using the ICPC-2, from 2,400 visits in another ED. Both EDs are responsible for primary care and special care urgent and emergent patients. An expert panel finalized the classification and the authors of this study implemented the new system in the Jorvi ED. This paper discusses the results.

This study created a structured complaint classification suitable for ED use and implemented it to practice. The classification is based on two stand-alone studies; one focused on identifying presenting complaints from data of 40,610 ED presentations and the other on a two-week survey while using the ICPC-2 as a classification for recording presenting complaints. The study produced a classification based on the ICPC-2, including 89 presenting complaints. The results from implementation were positive and encouraging and the classification is currently in routine use in the Jorvi ED.

Two major factors motivated the selection of the ICPC-2 as the foundation of the new classification. Firstly, the ICPC-2 is symptom-oriented. In the triage process, patients have no predefined diagnoses per se, and the triage nurse’s should base his/her interpretation of urgency and requisite resources on symptoms. Secondly, the ICPC-2 is compatible with the ICD-10. The ICPC-2 is a part of WHO’s Family of International Classifications (FIC) and all of the complaints can be traced back to related ICD-10 codes. However, the ICPC-2 has too much detail for effective use in EDs.

A short list of presenting complaints is simple and reliable, but if the list is too short, it does not present enough information. When the list is long, the specificity is higher but the system is complex and data analysis becomes difficult. With a long list, shortened modifications emerge and their use jeopardizes data comparison. ED-specific presenting complaint lists range between 33 [16] and 165 [12] in the number of different codes. The current study indicates that a list of 89 presenting complaints is well suited for ED use. In triage, time for recording codes is scarce; the system has to be easy to use. Obtaining presenting complaint information from a free text field is very time-consuming and not suitable for routine use.

Unlike the ICPC-2 based classification of this study, Canadian CEDIS classification is ED-specific. Although many presenting complaints are similar in both lists, there are remarkable differences. The main difference in this study’s shorter list is that the presenting complaints of many organs, such as ear, eye, or gynaecological symptoms, are not divided into as many subclasses as those in the CEDIS; patients in these subclasses have the same pathways and use the same resources in EDs. However, the study’s list included more subclasses to code traumas and symptoms of the extremities; symptoms of the ankle, knee, and hip have their own codes. Rare but urgent cases, such as periorbital oedema, were omitted because the inclusion of one such item would soon increase the list with other equally urgent, but rare, symptoms. One must keep in mind that almost all presenting complaints can include emergent cases.

In some cases the CEDIS list had more interpretation of the symptoms than did the list in the current study. The study list, for example, listed hyperventilation with dyspnoe to prevent premature conclusions of possible diagnoses. Hyperventilation can be a harmless symptom, but it can also be a symptom of ketoasidosis or pulmonary embolism. This same principle led to the decision to keep chest pain as one presenting complaint without trying to divide it into more or less specific cardiac features. This is important because a cardiac event is difficult to diagnose [23].

The ICPC-2 was quite easily modified and suites the classification of ED presenting complaints well. Many codes for the ED list were combined from two or three codes of the ICPC-2. However, some symptoms had to be modified from the ICPC-2. Hyperglycaemia is under abnormal investigation results in the ICPC-2, while hypoglycaemia has its own code and allergic reaction, which the new list had to include despite its features of diagnosis and the fact that it is coded as a diagnosis in the ICPC-2.

Classification of presenting complaints does not remove the need for using free text communication in EDs, it is highly important and codes or rules should not restrict its use. In the Jorvi Hospital ED, the presenting complaint code is not used for communication at all; it is registered only for purposes of data analysis and quality studies and free text is used for communication.

Decisions made in triage regarding urgency and tracks have significant effects on the duration and the quality of the care process. The presenting complaint is one of the most important variables affecting these decisions. The variety of patients entering EDs is wide, and to evaluate triage performance and care quality, divisions between different patient groups are necessary. Often studies concerning EDs make divisions and give treatment recommendations based on diagnoses or treatments but such information is not available at the point of triage. Presenting complaints provide more relevant divisions for studying quality, process, and outcomes of care; for such research, structured information regarding presenting complaints is essential.

In addition, routine research in the classification of presenting complaints in ERs enables several practical data usage possibilities. Presenting complaint information, for example, aids demand and capacity planning, streaming, and patient flow control, quality control, and benchmarking of performance.

The ICPC-2 classification can be easily modified for use in EDs by decreasing the number of codes. The use of structured classification for recording presenting complaints in EDs helps to compare and improve EDs, both nationally and internationally. Recording presenting complaints provides information from ED case-mixes and helps in the planning and control of patient flows.

It is essential that future research develops a similar, ICPC-2 based presenting complaint list for pediatric emergency care. Further validation studies should also be conducted to the classification in different environments using standard evaluation protocols such as COREQ [24].

The authors recommend the use of the ICPC-2 based list for recording ED presenting complaints; the new list with 89 presenting complaints was more convenient and easy to use. The classification is easily transferable to different EDs, although such a classification needs to be developed and improved continually. To ensure data comparability, all such improvement modifications should be centralized, either by national actor or by an international organization willing to take the classification forward, such as the WONCA International Classification Committee.