Key results
This population-based historic cohort study showed that patients with contact to two Danish emergency departments received a large variety of diagnoses within all ICD-10 chapters. As the EDs handle injuries as cuts, fractures and trauma, not surprisingly the most frequent chapter used was ‘Injuries and poisoning’. This accounted for more than one-third of all the contacts.
More interestingly, we found that almost one third of the contacts received a diagnosis within the two non-specific chapters ‘Symptoms and signs’ or ‘Other factors’. Although we looked for specific diagnoses in relation to all contacts, there was a large proportion of non-specific diagnoses. This is underlined by the single most used diagnosis being ‘Observation for suspected disease or condition, unspecified (DZ039)’.
Moreover, we found that 30-day mortality for the non-specific chapters surpassed several of the more organ-specific chapters. At day 1, the highest mortality was for ‘Circulatory diseases’ whereas the remaining chapters had a 1-day mortality below 2%.
The highest 30-day mortality was reported for ‘Neoplasms’. Chapters ‘Endocrine diseases’, ‘Respiratory diseases’ and ‘Circulatory diseases’ also had relatively high 30-day mortality.
We performed additional sensitivity analysis including only the first patient contact to estimate mortality (see Additional file 5: Table S5), which only lead to a slight change of order within the chapters with the highest mortality.
An interesting group of patients in the study were children < 1 years. This was the single most frequent age group in contact with the ED. Consequently, it would be interesting to investigate this group more thoroughly regarding diagnoses, mode of referral and parents’ perspective.
Strengths and weaknesses of the study
One major strength of the study is its population-based design. Thus, it includes all patient contacts to the EDs in this region, which minimizes selection bias and allows a strong follow-up.
Another major strength is the linkage of patient contacts to data registries through the unique civil registration number of each patient.
In the present study, we performed our analysis from the perspective of the ED, i.e. the main unit in our data was an ED contact. This meant the same patient could be included several times and the mortality estimates are therefore based on each patient contact. We decided on this setup, as this is the reality in an ED: every time the patient has a new contact, a new assessment of the patient is performed. We chose to complement this analysis with mortality estimates based only on the first contact the patient had to the ED, hence looking at it from the patient’s perspective. We considered it a strength, that we performed both analyses.
In the present study setup, it is not possible to elaborate on the reasons behind the non-specific diagnosis within the ED population. We do not know the extent of self-referred patients or the mode of referral in general. Similarly, information on GP use (daytime and out-of-hours) before and after ED contact and more extensive history of admissions and readmissions would contribute to understanding this patient group and the ED population in general.
The mortality within the non-specific patient group in the present study could be explained by comorbidity, which we did not investigate. With this, we would have been able to describe any differences in comorbidity between relevant patient groups e.g. high vs. low mortality or non-specific vs. specific diagnoses. Including this would strengthen future studies.
The mortality could also be explained by a number of life-threatening diagnoses within the two non-specific chapters – such as ‘respiratory arrest (DR092)’ and ‘cardiorespiratory failure (DR092A)’ albeit these represent very few cases (total n = 21). In the mortality rate estimation of the non-specific chapters, we excluded patient contacts with diagnoses concerning unspecified causes of mortality i.e. the patients are not alive at time of diagnosis (n = 122), as these cases would otherwise lead to an overestimation of mortality within the chapters.
Another minor weakness related to the mortality data could occur if a patient moved out of the region during the study. Data on vital status were limited to patients in the North and Central Denmark Region, which meant patients moving elsewhere and tourists, will have incomplete mortality reporting.
Other studies
Most international studies concerning patients in the ED revolve around patients with many contacts to the EDs and their characteristics [14, 15], whereas fewer studies have described the entire ED population. No study has described the emergency patient population regarding diagnosis and mortality in the new Danish ED setting.
A large part of ED studies originate from the US where the gatekeeper function of GPs is absent. This function aims to ensure that patients in need of more specialized care can access secondary healthcare facilities, but also that patients are guided to primary care, when this is sufficient. Some patients still by-pass this setup, a well-known issue in countries with similar healthcare setups [16, 17]. Considering the differences in setup, we primarily compared our study to studies from countries with similar healthcare organization e.g. Norway and Iceland.
Carter-Storch et al. published a Danish cross-sectional study based on data from 2010 [5] aiming to categorize complaints and symptoms of admitted ED patients into major groups. As a secondary result, they found the following distribution of patients: 49% medical, 31% surgical, 15% orthopaedic and 5% vascular surgical. This study included only admitted patients and did not use discharge diagnoses, but collected data regarding complaints and symptoms from the referring doctor or the patients, making it difficult to compare with the present study.
A Danish-American study by Dalgaard et al. [18] investigated and described the population in an ED in Boston, USA, and concluded that patients received non-specific diagnoses in 26.5% of the cases. The distribution of the remaining chapters were also in good agreement with our findings (albeit the classification used was ICD-9), which is interesting as the gatekeeper function of the GP was absent. Around two thirds of the patients presenting to the ED were discharged without admission - one could hypothesize that a part of these patients could have been seen by a GP instead.
In a Norwegian prospective study from 2014 [19], Langlo et al. assigned International Classification of Primary Care (ICPC-2) codes for presenting complaints and symptoms of 3163 patients in an ED during a 2 month period. They found the most frequent ICPC-2 classifications chapters to be general and unspecified 37%, digestive 19%, respiratory 12%, neurological 12% and musculoskeletal 6%. Although, the ICPC-2 and ICD-10 classifications differ, there are some similarities. Non-specific chapters also account for almost one third in our study and diagnoses of digestive and respiratory diseases are likewise quite frequently given. During daytime, the Norwegian ED has a separate clinic handling injuries and minor traumas, which explains the dissimilarity in the findings.
Vest-Hansen et al. investigated 264,265 acute hospital admissions to medical wards (not EDs) during 2010 in a large Danish population-based observational study [20] and described the pattern of diagnosis for this group. The most frequent ICD-10 chapters were non-specific (28.7%), circulatory diseases (19.3%), infections (15.5%) and injuries and poisoning (6.3%) (adverse effects, intracranial injury and poisoning by psychotropic drugs were the most frequent diagnoses in this chapter). The study included only patients admitted to medical wards, which could explain the higher proportion of circulatory diseases and infections, whereas orthopaedic patients and other surgical patients were not included.
Perhaps the non-specific patient group needs further attention. This is somewhat underlined by a Danish study where Hansen et al. [21] followed 409 patients with initial non-specific diagnoses (DZ03) in an acute medical admission unit. In the patient group discharged with non-specific diagnoses, 28% were readmitted within 30 days of discharge and 76% of these received a more specific diagnosis.
An interesting American study from 2014 by Raven et al. [22] investigated and found only very limited correspondence between discharge diagnoses from the ED compared to the patients’ presenting complaints in the ED. However, this study used ICD-9, which is less updated and contains much fewer diagnoses than ICD-10 [23], which could have meant that the diagnoses used were less specific. In addition, the settings for these studies were quite different. The Danish setting had a gatekeeper function in the form of a GP (both during work hours and after hours). Thus, many of the ED patients have been assessed by a GP before their ED contact, which could have led to a better correspondence between complaints and discharge diagnoses.
In an Icelandic descriptive study from 2006 by Gunnarsdottir et al. [24] investigated patients discharged from the ED without admission. Patients within the psychiatric, paediatric, gynaecology and obstetrics fields were not included. The study included the years 1995–2001. In the last year included, they too found the non-specific ICD-10 chapters to account for 29.4% of the contacts followed by circulatory diseases (16.5%) and diseases of the genitourinary tract (10.8%).
One of the key result of the present study is the mortality within the non-specific chapters. In a recent study of 148,757 patients from the same Danish region, Christensen et al. [6] found that patients receiving non-specific diagnoses after being brought to hospital by ambulance had a 30-day mortality of 4.3%. This correlates well with our findings as did the distribution of the most frequent ICD-10 diagnosis chapters (injury and poisoning 30.0%, symptoms and abnormal findings, not elsewhere classified 17.5% and factors influencing health status and contact with health services 14.1%).
Interpretation
This study shows that ED patients receive a large variety of diagnoses across all ICD-10 chapters, but the majority of patient contacts leads to a diagnosis within the ‘injuries and poisoning’ or non-specific ICD-10 diagnosis chapters. Similar large proportions of non-specific diagnoses have been found in various other studies from comparable healthcare settings.
Several reasons could be at play as to why we find this large group of non-specific diagnoses. On one hand, we see an increasing overload of the EDs, less available beds for admissions and consequential demand for handling more patients in the ED, which could lead to more use of non-specific diagnoses. On the other hand, a large proportion of the patient population is characterized by chronic disease and multi-morbidity leading to complex symptomatology that does not fit well in the confinement of the diagnoses of the ICD-10.
Only very few studies describing the ED populations exist and the results of this first description of a Danish ED population’s diagnosis and mortality could interest public healthcare planners and policy experts, when making administrative decisions.
Moreover, the Danish EDs have changed in recent years and this study elucidates some of the implications following the reorganization i.e. changes in the patient population.
Taken into consideration that a Danish emergency medicine medical specialty has emerged perhaps the education planners may find the study relevant.