Article Text
PDF
Abstract
Objective It is often asserted that the crowding phenomenon in emergency departments (ED) can be explained by an increase in visits considered as non-urgent. The aim of our study was to quantify the increase in ED visit rates and to determine whether this increase was explained by non-severe visit types.
Methods This observational study covers all ED visits between 2002 and 2015 by adult inhabitants of the Midi-Pyrénées region in France. Their characteristics were collected from the emergency visit summaries. We modelled the visit rates per year using linear regression models, and an increase was considered significant when the 95% CIs did not include zero. The severity of the patients’ condition during ED visit was determined through the ‘Clinical Classification of Emergency’ score. Non-severe visits were those where the patient was stable, and the physician deemed no intervention necessary. Intermediate-severity visits concerned patients who were stable but requiring diagnostic or therapeutic procedures.
Results The 37 studied EDs managed >7 million visits between 2002 and 2015. There was an average increase of +4.83 (95% CI 4.33 to 5.32) visits per 1000 inhabitants each year. The increase in non-severe visit types was +0.88 (95% CI 0.42 to 1.34) per 1000 inhabitants, while the increase in intermediate-severity visit types was +3.26 (95% CI 2.62 to 3.91) per 1000 inhabitants. This increase affected all age groups and all sexes.
Discussion It appears that the increase in ED use is not based on an increase in non-severe visit types, with a greater impact of intermediate-severity visit types requiring diagnostic or therapeutic procedures in ED.
- access to care
- emergency care systems, admission avoidance
- emergency care systems, primary care
- emergency department utilisation
- epidemiology
Statistics from Altmetric.com
- access to care
- emergency care systems, admission avoidance
- emergency care systems, primary care
- emergency department utilisation
- epidemiology
Key messages
What is already known on this subject
Over the last 30 years, many countries have faced emergency department (ED) crowding.
This crowding has been, at least in part, attributed to an increase in ED attendance for reasons considered as ‘non-urgent’.
The role of non-severe visits in ED crowding is, however, still controversial.
What this study adds
This study provides some insight into the involvement of severity visit types in the crowding phenomenon in the French care system, based on data of the longest running French regional ‘emergency department observatory’.
Our results confirmed an increase in ED visits unexplained by population growth nor by the population aging.
This increase in ED visit rates was mainly based on an increase in intermediate-severity visit types, that is, non-admitted visits but requiring diagnostic or therapeutic procedures.
Introduction
Over the last 30 years, many countries have had to face the crisis of emergency department (ED) crowding,1 a phenomenon which impacts on treatment delays and mortality.2 In the medical literature, as well as the media, this crowding has been, at least in part, attributed to an increase in ED attendance for reasons considered as non-urgent.2 However, the question of the role of non-severe visits in ED crowding is still controversial.3
In France, health coverage is publicly funded and supports all types of primary care use, which can be freely chosen by patients. This care system is therefore theoretically robust to manage for ED-alternative care in non-severe cases. Yet, France also faces the problem of ED crowding.1 The 2015 Grall Report, which reviewed the change in ED care delivery since major reorganisation occurred in the 1990s, found that ED visit rates had doubled from 125 visits per 1000 inhabitants in 19934 to 286 per 1000 inhabitants in 2013.5 In the early 1990s, this increase had already been attributed to visits considered as ‘inappropriate’.4 In 2000, a published study by the Ministry-of-Health Statistics Department (DREES) had also linked the rising ED attendance with the increase in visits for emergencies known as ‘(only-) perceived emergencies’, that is, visits which did not require immediate care according to the professionals.6 Since then there have been several other attempts to quantify and prevent ‘non-severe’, ‘non-urgent’, ‘low-acuity’, ‘preventable’ or ’inappropriate' visits to French EDs, based on the assumption that their contribution to the crowding phenomenon was proven.7–9 However, no published study has, to our knowledge, analysed the temporal change in the severity of visits to the ED.
The investigation of both the evolution of ED visit rates and severity of visits to the ED may provide some insight into the involvement of these factors in the crowding phenomenon. The French situation allows us to study the evolution of ED usage over recent years independently of major financial barriers to alternative primary care and explore the question of whether people are more prone to non-severe ED visits when general practitioner (GP) care is theoretically financed in the same way as ED care. The longest running French regional ‘ED observatory’, established in 2001 in the Midi-Pyrénées region, provides a real opportunity to explore this question. This register has routinely and exhaustively collected data from all the public and private EDs of the largest French region over the last 14 years. These data therefore allow us to meet both of our objectives: (1) to quantify the increase in ED visit rates among our population and (2) to determine whether this increase was explained by non-severe visit types.
Material and methods
Setting
The Midi-Pyrénées region has almost 3 million inhabitants, with 1.3 million living in Toulouse and its urban metropolitan area (the fourth largest in France). The rest of the region is predominantly rural. Before the recent restructuring of French region (2016), the Midi-Pyrénées region was the largest in France. The health insurance and health service organisations are centralised at the national level, so the region is representative on these aspects of the French system overall.
Data sources
We examined data from the ED observatory database for the French Midi-Pyrénées region, the oldest register of ED visits in France. This observatory was established in 2001 to exhaustively collect data from ‘emergency visit summaries’: a standardised coded form, filled in by each ED. These contain demographic, diagnostic and procedural data, routinely and exhaustively collected by private and public hospitals (covered in the same way by health assurance). We used the emergency visit summaries for all visits made between 2002 and 2015 to all EDs in the Midi-Pyrénées region. For the calculation of rates, we used the census data.
Study population
All adult visits to EDs in the register made between 1 January 2002 and 31 December 2015 were eligible for inclusion. We limited our analysis to inhabitants aged 15 years and over, because of the differences in ED usage patterns between adult and paediatric populations. Because of material constraints, we dealt with a 10% random sample (random draw by a uniform law, without stratification, followed by a check of the sample representativeness for the main variables), and the rates were calculated after a factor 10 correction to take account of this sampling.
Determination of severity
To measure the severity of the visit, we used the severity-level recorded by the French Clinical Classification Cf emergency (CCE) score, where a higher score is more severe (see figure 1).10 This score, based on the patient’s state on arrival, is coded at the end of the passage by the emergency physician. The non-severe visit types were defined as visits for which the physician considered, after his or her clinical examination, that the patient did not need any therapeutic or diagnostic procedure in the ED (CCE-1). By comparison, intermediate-severity visit types were defined as visits for which the physician considered that the patient’s clinical state was ‘stable’ (ie, not life-threatening and not likely to deteriorate) on his or her arrival, but for which at least one therapeutic or diagnostic procedure was performed at ED (CCE-2).
Clinical classification of emergencies (CCE).
On an exploratory basis, the outcome (admitted or non-admitted), the reason for visit and the diagnosis were also analysed. The ‘reason for visit’ was based on the type of patients’ initial complaint (medicosurgical, traumatic, psychiatric or other) and also coded by the emergency physician. The ‘diagnosis categories’ was based on the physicians’ final and main diagnosis, coded with the ICD-10 (International Classification of Diseases) by the physician and grouped by medical discipline by a national standardised algorithm of the observatory.
Statistical analysis
To quantify the increase of ED visit rates, we first analysed the change in the annual number of visits, the change in the annual visit rates for 1000 inhabitants and, finally, the change in the annual standardised visit rates for 1000 habitants. To calculate these rates, the reference populations used were the regional populations over 15 years of age for each analysed year. For the standardised rates, we weighted the visit rates for each age group (in 5 years’ increments) and sex by the age and sex structure of the 2002 population, that is, the population of the first year of this study. The standardised rates therefore corresponded to the annual visit rates as if the population structure had remained stable since 2002. To measure the increase, we modelled the gross visit rates per year with linear regression models, after having graphically verified the linearity of the change over time. The regression coefficient β therefore corresponded to the increase in visit rates for an increment of 1 year, that is, the ‘Number of Additional Visits per 1000 inhabitants each Year’. An increase was considered as significant when the 95% CI of the coefficient did not include zero.11 Some variables used for the analysis had missing data and were imputed with the Multiple Imputation using Chained Equation protocol.12 The variables relating to the type of institution, the year and the day of the visit, which did not have any missing data, were also used for the imputation. Data on diagnoses were more precise than data relating to the reason for visit but contained 12% of missing data, which were not imputed due to the large number of diagnoses and the heterogeneity of the diagnostic categories. Diagnostic categories were therefore analysed on patients with complete data.
Ethical approval
As a non-interventional study based on retrospective data, this study did not require the agreement of the Comité de Protection des Personnes under French law. Concerning data protection, the data collection and analysis were in compliance with French regulations.
Results
Description of visits
The regional ED observatory recorded the 7 million visits managed by the 37 Midi-Pyrénées EDs between 2002 and 2015. The study therefore dealt with 14 annual visit rates. The median age was ~45 years, and patients were more likely to be male. The main reasons for visits were medicosurgical and traumatic. Most visits were coded as intermediate (CCE-2). Most patients (71.64%) returned home. Details are given in table 1.
Description of complete and imputed data in the 10% sample (n=721 034)
Evolution of ED visit rates
The annual number of ED visits increased over the studied 14-year period, with >220 000 additional visits between 2002 and 2015 (ie, up from 401 820 to 624 010), corresponding to an increase of +55%. This increase was only partly explained by the increase in the regional population (33%), and marginally by the change in the population structure (3%). Details are given in table 2.
Evolution of annual ED visit rates
Graphically speaking (see figure 2), the evolution of visits to the ED since 2002 was linear, allowing for the use of a linear regression model. This model confirmed a significant increase in ED visit rates of +4.83 (95% CI 4.33 to 5.32) for an increment of 1 year.
Evolution of emergency departments visits rates between 2002 and 2015, by severity
Evolution of ED visits characteristics
Table 3 describes the evolution of the visit rates for each characteristic. The annual visit rates with intermediate severity have increased by +3.3 per 1000 inhabitants every year since 2002, while non-severe visit rates and severe visit rates have only increased by +0.8 and +0.9 each year, respectively. The increase in ED visit rates and particularly in intermediate-severity visit rates were mainly due to non-admitted visits. We also observed that almost the only visits to have actually increased were medicosurgical visits, compared with the traumatic, psychiatric or other visits. No diagnostic category alone explained the increase in visit rates, although visits for digestive reasons had increased slightly more than the others. The day and time of the visits does not seem to have changed since 2002. Indeed, there was an increase in visit rates across the entirety of the week and of the day.
Characteristics of the evolution of visit rates
This increase in visits was found for all age groups and all sexes, with a steeper slope (increase) for young women and people over 75 years of age (table 3). For people over 75 years of age, there was increase of +9.4 visits each year per 1000 inhabitants (see table 4). Among this population, the increase in non-severe visit types was +0.4 while the increase in intermediate-severity visit types was +5.9 and +2.9 in severe visit types. Both admitted and non-admitted visits rose: respectively, +3.1 and +6.4 visits per 1000 inhabitants per year. Across the whole population, the age-standardised and sex-standardised visit rates did not differ from the gross rates (see table 2), suggesting that the change in the population structure did not explain the increase in visit rates. But the evolution of the structure in the over 75 s group could have an impact. Indeed, there was a relative increase in the frequency of very elderly people (over 85 years and even over 90 years). This ageing of the elderly population partly explains the increase in visit rates in this group, as the evolution of the standardised rates was of +7.1 visits for 1000 inhabitants over 75 per year, against +9.4 with the gross rates.
Characteristics of the evolution of visit rates in people over 75 years of age
Discussion
Main findings
Our results confirmed an increase in ED visits unexplained by population growth. Indeed, every year, each 1000 persons produced 4.8 more visits to ED, that is, ~12,000 more visits per year for a population of 2.4 million people. The increase in non-severe visit types was +0.88 (95% CI 0.42 to 1.34) while the increase in intermediate-severity visit types (made by stable patients but requiring diagnostic or therapeutic procedures) was +3.26 (95% CI 2.62 to 3.91).
Methodological strengths and limitations
The strengths of this study include the length of its duration and, above all, the spatial exhaustiveness of the data collection. These points allowed us to study a multicentre regional network rather than several isolated institutions, as is often the case in this kind of study.2 Data were collected routinely, thoroughly and in a standardised way since 2002, so the increase in visit rates cannot be attributed to a collection bias. Neither can it be attributed to seasonal nor daily variation as is the case for studies performed over a single given day.
The main limitation may be our choice of definition of non-severe visit. This concept is complex and does not allow for a single definition.13 Moreover, in this field, it frequently refers to visits that could have been managed by a GP. Even if it has already been done,14 relying on the performance of a diagnostic or therapeutic procedure to assess severity (or non-avoidability) of ED visits is not classical.13 However, the CCE score is largely used in France in this field,8 9 15 for the main reason that it is systematically collected by EDs and well known by the French physicians. Its limitations are known, in particular the fact that it is difficult to conclude about the avoidability or non-avoidability of visits classified as CCE-2.16
We chose to define non-severe visits as visits which were coded ‘CCE-1’ as we considered the realisation of a complementary procedure as a kind of legitimation of visits. Indeed, if treatment has been administered in the ED, it may be deemed that a prescription for delivery by a pharmacy would not have been enough. Similarly, a diagnostic procedure (biology, imaging and so on) performed by the ED physician reflects the need to eliminate at least some severe diagnoses before returning the patient to his or her home. But it may be objected that these procedures can be carried out by opportunity, in compliance with protocol or to protect the institution, and therefore that some of these visits could be considered as non-severe. Our results show that non-medical visits do not seem to be involved in the increase of ED visits because visit rates for ‘administrative, forensic or social reasons’ are stable and because the increase in visits which do not require therapeutic or diagnostic procedures (CCE-1) was negligible.
Visits mainly implicated in the increase are visits made ‘by a patient whose clinical condition is stable and who requires at least one complementary procedure’ (CCE-2). However, the question remains as to whether these types of visits actually required management by EDs, notably because the level-2 of the CCE score must be carefully interpreted. First, the distinction between intermediate-severity and severe visit types was based on the clinical stability at arrival but was coded at the end of the passage. This might therefore under-estimate the level of concern the patient or physician might have in regard to the patient’s evolution during the stay. Then, the CCE-2 visits are highly heterogeneous situations, involving >60% of visits. Some study has considered all CCE-2 visits as avoidable,8 9 but it is rather recognised today that, even if some of them can be avoidable, it is not the majority.16 It reinforced our choice to use only the CCE-1 category to define non-severe visits.
However, this question does not allow for a methodological or statistical answer. It primarily depends on the perceived role of EDs within the context of primary care services. A minimalist position regarding its role can be defended, where the EDs are only designed to provide care for acute and serious pathologies which may affect the vital prognosis of the patient. This is a concept close to the historical role of the contemporary ED, and ED workers are primarily trained for this kind of practice.5 In this case, the visit types we highlighted that are driving the increase in visit rates would not be deemed suitable for ED management. On the contrary, it is possible to defend an inverse and more extensive conception which considers emergency services as a primary care service like any other. In such cases, all visits are deemed suitable for ED management, although some could also have been managed by a GP. It remains difficult to measure the number of these visits, and also to identify the share of visits that have slipped from one service to another. A study of the parallel evolution of the use of GPs would allow us to better understand this evolution.
Comparison with other studies
Several countries are affected by increasing ED visit rates,1 but without the same basal rates, nor the same increase. In the USA, for example, the visit rates are higher, but have increased more slowly with only around 2–3 additional visits per 1000 inhabitants each year: 389 annual visits per 1000 inhabitants in 2002 against 419 in 2013.17 The UK has faced an increase of around 7–8 additional visits per 1000 inhabitants each year: 240 annual visits per 1000 inhabitants in 2002 against approximately 350 in 2015.18 The increasing use of EDs by the population therefore seems to be a cause of ED crowding, along with other factors such as inadequate staffing or hospital occupancy levels.2 More and more people are going to the ED, but this does not mean that there are more and more non-appropriate visits. No other published study has, to our knowledge, analysed the year by year rate of ED visits by severity.
Conclusion and implication
Since at least 2002, there has been an increase in visits to the ED, independently of population size increase or structural changes. It appears that the increase is not attributable to non-severe visits but rather to moderately severe visits. These findings are observed within a country which has a robust ED-alternative care system, theoretically covered in the same way as ED. This calls into question the place of EDs in the primary care organisation, and even the primary care organisation itself, in order to adapt to a change in usage.
Abstract translation
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.Footnotes
Contributors HC wrote the statistical analysis plan, cleaned and analysed the data, drafted, wrote and revised the paper. She is guarantor. FP and LP monitored data collection and revised the paper. TL, MK-H and SC initiated the collaborative project, revised the statistical analysis plan and revised the paper. OA initiated the collaborative project, monitored data collection and revised the paper. SL wrote the statistical analysis plan, analysed the data, drafted and revised the paper.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Ethics approval The Midi-Pyrénées ED Observatory database was declared to the French National Commission for Data Protection and Liberties (CNIL) under declaration no 761 633 (31/08/2001). The database was anonymous.
Provenance and peer review Not commissioned; externally peer reviewed.
Patient consent for publication Not required.