1 Faculdade de Medicina, Universidade Federal de

This systematic review aimed to measure the prevalence of inappropriate emergency department (ED) use by adults and associated factors. The review included 31 articles published in the last 12 years. Prevalence of inappropriate ED use varied from 20 to 40% and was associated with age and income. Female patients, those without co-morbidities, without a regular physician, without a regular source of care, and those not referred to the ED by a physician also showed more inappropriate ED use, with the relative risk varying from 1.12 to 2.42. Difficulties in accessing primary health care (difficulties in setting appointments, longer waiting periods, and short business hours at the primary health care service) were also associated with inappropriate ED use. Thus, primary care requires fully qualified patient reception and efficient triage to promptly attend cases that cannot wait. It is also necessary to orient the population on situations in which they should go to the ED and on the disadvantages of consulting the ED when the case is not really urgent.


Inappropriate use of emergency services: a systematic review of prevalence and associated factors
Prevalência e fatores associados ao uso inadequado do serviço de emergência: uma revisão sistemática da literatura

Introduction
Emergency treatment aims to perform procedures to immediately relieve well-circumscribed situations, and is not intended to include on-going care 1 .However, patients frequently seek the emergency department (ED) to obtain immediate attention in order to perform tests and administer medication to relieve symptoms.Although this may appear appropriate from the patient's perspective (given existing limitations in other levels of health care), this type of use places a burden on the health system and increases the demand on the ED for care that could be managed better at other levels and that in a sense competes with true emergency cases.Meanwhile, such demand generates a kind of care that fails to create a bond with the health service (in which patients would receive not only treatment to relieve their immediate symptoms, but also health education) or to link with on-going care in order to prevent complications and new illnesses 2,3,4 .Such use of ED services is thus considered inappropriate.
The inappropriate use of ED services is a common problem in various countries, and the issue has thus been studied for more than two decades 5 .However, the wide range of criteria used to define inappropriate use of ED services has been an obstacle to systematic reviews on the topic 6 .Numerous studies have evaluated its prevalence and many have also examined factors associated with such inappropriate use, but most studies have been limited to evaluating sociodemographic factors, without examining other aspects in depth, such as those related to other levels of care.
The existing systematic reviews approach the historical profile of inappropriate ED use in the United States 7 , criteria to define inadequacy of ED services 8 , patterns in ED use by the elderly 9 , and the role of expanding primary health care and bad experiences in primary health care in determining inappropriate ED use, in publications from 1971 to 1997 10 .However, despite the existing reviews, there are no recent articles summarizing the findings on prevalence of inappropriate ED use or its determinants, particularly those related to health needs and access to health services.Thus, the current article presents a systematic literature review, exploring the principal criteria used to define inappropriate ED use, identifying the problem's prevalence and examining the state-of-the-art on factors associated with inappropriate ED use.Knowledge of the prevalence and factors associated with inappropriate ED use can help orient public policies to reduce the problem.

Methods
A literature review was performed in the following databases: MEDLINE (International Health Sciences Literature), LILACS (Latin American and Caribbean Health Sciences Literature), SciELO (Scientific Electronic Library Online), Harvard Electronic Library, website of the World Health Organization (http://www.who.int),website of Free Medical Journals (http://www.freemedicaljournals.com),and the Library of the Graduate Course in Epidemiology at the Federal University in Pelotas, Brazil.In addition, all the references from the initially selected articles were reviewed.Inclusion criteria for the articles were: indexed articles, published since 1995, with samples of adults, measuring prevalence and/or factors associated with inappropriate use of emergency services.Articles were excluded that did not clearly present the criteria used to define the outcome.
The following words were searched in "all fields": inappropriate, inadequate, nonurgent, non-urgent, misuse, appropriateness, unnecessary, nonemergency, demand, visits, consultation AND emergency department, emergency service, emergency services, emergency room, urgency department, urgency service, urgency services, urgency room, and their equivalents in Spanish and Portuguese.

Qualitative evaluation
The selected articles that evaluated factors associated with inappropriate ED use were classified according to the criteria proposed by Downs & Black 11 .Since no experimental study was selected, the items that only applied to this type of study were excluded.Using this criterion, we evaluated whether: • the study's hypotheses and objectives were described; • the principal outcomes to be measured were described; • the characteristics of the subjects included in the study were described (cohort or case-control); • the distribution of the principal confounding factors in each comparison group was described; • the study's principal findings were described; • in cohort studies, the characteristics of the subjects lost to follow-up were described; • the probability of association (p-value) with the principal outcomes was reported precisely; • the sample selected for the study was representative; • the sample included in the study was representative; • analyses not originally planned in the project were clearly indicated; • in cohort studies, the different follow-up times were adjusted in the analysis; or, in case-control studies, the time transpired between determinants and outcome was the same for cases and controls; • the statistical tests used to measure the principal outcomes were appropriate; • the measurements to evaluate the principal outcomes were accurate; • the individuals in the different groups were recruited from the same population; • the individuals in the different groups were recruited during the same time period; • the principal confounding factors were adequately adjusted in the analysis; • losses of individuals during follow-up were considered; and • the study's power to detect important effects was sufficient, with a significance level of 5%.
Two epidemiologists performed the scores, with a total of 18 items assessed according to the Downs & Black criteria, with a maximum score of 19 points.Whenever there was disagreement, the two epidemiologists conferred until reaching a consensus.

Data extraction and synthesis
Data were extracted from the studies with the respective year of publication, country where the study was performed, and design.Evaluation of factors associated with inappropriate ED use was based on the Downs & Black score, study characteristics, and effect measurements, analyzing the consistency of the results.

Results
A total of 5,124 articles were identified in the online databases, of which 4,938 were ruled out by reading the title and abstract, since, since they did not specify the study outcome.Of the 186 remaining studies, 31 were selected that evaluated prevalence rates and 22 that measured factors associated with inappropriate ED use.

Criteria for defining inappropriate ED use
The criteria for defining an inappropriate ED visit varied considerably between the studies, and the following were the most frequent: waiting time until receiving care, medical assessment of risk of death or target organ injury, and resources needed for care (whether only available in the ED or not).Other aspects included: self-perceived urgency, need for diagnostic tests, immediate treatment, and/or observation, and others (Table 1).Each of these items also varied greatly.For example, the list of tests considered in the definition of urgencies varied for each criterion.
We identified two standardized criteria for defining inappropriate use: (i) the Canadian Triage and Acuity Scale (CTAS) 12 , which classifies ED care in five levels, based on the waiting time allowed for the patient to be examined by the physician, risk of death, vital signs, pain level, possibility of complications, and origin of the injury and (ii) the Hospital Urgencies Appropriateness Protocol (PAUH) 13,14,15,16 , defining the case as urgent whenever one of the items evaluated was met.This protocol considers criteria of severity (for example, altered vital signs and active hemorrhages), diagnosis (like orders for laboratory or imaging tests), treatment (like need for IV medication), and origin of the need for seeking the ED (for example: coming directly from a traffic accident or symptoms suggesting vital urgency) (Table 1).

Prevalence
Prevalence of inappropriate ED use varied from 10 to 90%, based on the criterion used, and in nearly half of the studies it varied from 24 to 40% 6,12,13,14,16,17,18,19,20,21,22,23,24,25,26 .All these articles presented similar criteria, evaluating the number of hours the patient could wait without risk of death, need for tests or treatments, need for hospitalization, possibility of treatment at another level of care, and observation time (Table 1).
Of the 11 articles that reported prevalence rates for inappropriate ED use greater than or equal to 45%, seven 6,15,27,28,29,30,31,32 used more rigorous criteria to define appropriate ED use, such as requiring observation for more than 12 hours or hospitalization; three other 33,34,35 used samples that excluded populations presenting greater severity.Meanwhile, among seven studies 6,36,37,38,39,40,41 that found prevalence rates lower than 20%, three used self-perceived urgency as the definition of appropriate ED use 6,37,38 ; two others used more rigorous criteria to define inappropriate use 36,41 ; and the study by Finn et al. 40 evaluated a lower-risk population for inappropriate use (65 years or older) (Table 1).

Characterization of studies on association
The review included 19 cross-sectional studies, two case-controls, and one cohort.Eight studies conducted were in European countries, seven in North America, one in Central America, one in Oceania, three in China, and two in South America.
The methodological score varied from 5 to 13 points, with a mean of 9 (median = 8).Of the studies evaluated, five had a score or 7 or less and eight had scores of 10 or more.In addition to the limitations related to study design (six items could not be scored in 19 studies, since they were cross-sectional), the principal limitations in the Downs & Black classification were those related to data analysis, such as performing only univariate analysis, and the failure to describe the strategy used in the multivariate analysis, or to the sample's representativeness.Table 2 shows the scores for each study.

Characteristics of inappropriate ED use
As for time of arrival at the ED, the early morning hours (midnight to 08:00 AM) was the period with the most appropriate ED use 13,14,21,22,23,28 , while the morning and afternoon shifts showed the heaviest ED use by patients with inappropriate complaints 13,21 .Day of the week was not associated with inappropriate use 12,14,22,37 .After controlling for confounding factors, duration of symptoms was directly associated with inappropriate ED use 12,13,22,37 .Derlet et al. 41 Gill & Riley 38 Gill et al. 29 Rubin & Bonnin 24 Petersen et al. 34 Young et al. 35 Criterion Liu et al. 30 Sarver et al. 25 Billings et al. 18 Michelen et al. 20 McCaig & Nawar 39 Afilalo et al. 36 Béland et al. 17 • Urgent: when patient required immediate attention, with risk of death or functional injuries.
• Non-urgent: other situations Classified as urgent or non-urgent:

Europe
Dale et al. 19 Coleman et al. 33 Oterino et al. 14 Sempere-Selva et al. 16 Sánchez-López et al. 15 Pereira et al. 22  Classified in 5 levels of severity, based on waiting time to be evaluated, frequency with which patient could be evaluated while waiting to be seen, and presence of complication (altered vital signs, level of pain, and mechanism of injury, e.g., car accident).

CTAS
• Level 5 (Non-urgent): patient could wait 2 hours to be seen by MD without risk of death and could be investigated or receive intervention elsewhere in the health system Classified as true urgency or not: • True urgency: after medical evaluation, the patient presented indication of remaining in hospital for 24 hours, need for surgery, ICU or death.
• Non-urgent: other situations Classified as urgent or non-urgent: • Non-urgent: came by themselves, could be resolved by primary care, no need for immediate intervention.
Classified as urgent or non-urgent: • Non-urgent: patients with non-urgent injuries or diseases that could be treated elsewhere.
Classified as true urgency or not: • Urgent: criteria for severity of signs and symptoms, treatment, diagnoses, related to follow-up on consultation and extra criteria for patient not referred by MD.
• Non-urgent: did not meet any of the above criteria.
Classified as urgent or not: • Urgent: criteria for severity of signs and symptoms, treatment, diagnoses, related to follow-up on consultation and extra criteria for patient not referred by MD.
• Non-urgent: when did not meet any of the above criteria.
Classified as urgent or not: • Urgent: criteria for severity of signs and symptoms, treatment, diagnoses (excluded X-ray), related to follow-up on consultation and extra criteria for patient not referred by MD.
• Non-urgent: did not meet any of the above criteria.
Classified as appropriate or not:  21 Bianco et al. 37 Asia Shah et al. 31 Lee et al. 32

Others
Rodríguez et al. 23 Dent et al. 27 Finn et al. 40 Stein et al. 26 Based on CTAS • Non-urgent: patient that could be evaluated in the ED or elsewhere after more than 6 hours without risk of death.
Classified in 4 levels: extreme emergency, emergency, emergency as perceived by patient, and non-urgent.
• Non-urgent: patient with no acute symptoms or the latter were recent and minor, without self-perceived emergency, wishing to have a check-up, renew a prescription, or excuse to return to work.
• Non-urgent: in MD's assessment, did not present risk of death or did not require care within hours to decrease discomfort.
Classified as true urgency or not: • Non-urgent: could be managed as outpatients by GP, could wait several hours to be seen without risk of death or harm to patient.
Classified in 4 levels: first, second, and third priority urgency, and non-urgent.
• Non-urgent: chronic disease problems, not acute-onchronic, no risk of death in their evolution, could wait 24-48 hours to be treated by family doctor.
Classified as appropriate or inappropriate: • Appropriate: when referred by health professional classified from 1 to 3 on the Australian scale, required ≥ 4 hours of treatment or observation, was hospitalized or died, arrived between 10:00PM and 07:00AM.
• Inappropriate (treatable by GP): other situations Classified as appropriate or not: • Appropriate: could not be evaluated or managed by primary care or at patient's home.By definition, whenever patient was hospitalized, consider appropriate.
• Inappropriate: all other situations Classified in 3 levels: emergency, urgent, and non-urgent.
• Elective consultation in ED: when could consult in > 24 hours.• Urgent: criteria for severity of signs and symptoms, treatment, diagnoses, related to follow-up of consultation and extra criteria for patient not referred by MD.
• Non-urgent: did not meet any of the above criteria.
The three studies that evaluated the association between means of transportation 12,16,26 and inappropriate use found varied results, which can be explained by the variation in each site's characteristics (size of the city, population's socioeconomic status, quality of mass transportation, and availability of ambulance services, among others).The majority of the studies that investigated proximity to the ED did not show a significant association with inappropriate use 14,16,32,37 .However, according to Oktay et al. 21, proximity to the ED was the principal reason for inappropriate complaints.Likewise, Young et al. 35 found that distance was one of the barriers to utilization of primary health care by individuals that visited the ED inappropriately.

Demographic variables
Sixteen studies showed an inverse association between age and inappropriate ED use 12,13,14,16,19,21,22,23,25,30,31,32,33,34,37,42 .Only two studies 40,43 failed to show this association.In the first, the study population consisted of adults 65 years or older; the second, in addition to only comparing mean age in two groups, showed a lower score based on the Downs & Black criteria.
The association with gender was evaluated in 16 studies, eight of which 13,14,21,22,25,30,34,37 observed (in the multivariate analysis) that women showed higher odds of inappropriately visiting the ED.In the studies, the RR varied from 1.12 to 1.56.Carret et al. 13 stratified their sample in two age groups and only found this association in the younger groups (15-49 years) (RR = 1.52, p < 0.001).
Among those not showing an association with gender 12,16,19,23,31,32,40,43 , five 16,19,23,31,43 only performed univariate analysis and/or showed a low score according to Downs & Black criteria and Finn et al only studied individuals 65 years or older 40 .Two other studies that did not show an association with gender were case-controls; however, they compared populations with similar health problems that chose to go to the ED or to the outpatient department, while the others compared individuals with inappropriate versus appropriate ED use 12,32 .The study by Shah et al. 31 was performed in Kuwait, where important cultural differences could explain this lack of association.
Of six studies that evaluated marital status 12,13,16,31,34,37 , only Afilalo et al. 12 showed an association between living alone and appropriate ED use (p = 0.046).Five studies examined race or skin color, but since they categorized it differently it was impossible to summarize the results 13,25,30,34,43 .

Socioeconomic variables
Three studies found a direct association between socioeconomic status (schooling 22,31 and income 21 ) and inappropriate ED use 21,22,31 .Bianco et al. 37 showed a borderline direct association between schooling and inappropriate ED use (p = 0.052), as did Carret et al. 13 for the group 50 years or older (p = 0.06).Lee et al. 32 found no association between schooling and inappropriate use, but individuals that owned their own homes used the ED more inappropriately than those that consulted primary care for similar complaints.
Three studies found no association between socioeconomic status and the outcome 12,26,34 , but Petersen et al. 34 compared groups with more (versus less) than 12 years of schooling and Afilalo et al. 12 only performed a univariate analysis for this association.Meanwhile, Sarver et al. 25 , in a cohort study of a population that reported having a regular non-emergency source of care showed an inverse relationship between income and inappropriate ED use.
In most of the studies, occupational status was not associated with inappropriate ED use  (continues) 12,13,25,34,43 .Lee et al. 32 showed protection against inappropriate use for individuals that worked part-time and for housewives, compared to those that worked fulltime.Pereira et al. 22 observed that retired women visited the ED 70% more appropriately than wage-earning women, even after adjusting for age.

Health needs
Some authors have reported that individuals without chronic non-communicable diseases or co-morbidities visit the ED more inappropriately 12,14,40 , while others have found the same in a specific age group (50 years and older) 13 or in persons without specific co-morbidities (cardiac and pulmonary) 34 .The magnitude of the asso-   ciation varied from 1.5 to 2.0.Meanwhile, Bianco et al. 37 and Carret et al. 13 , among adults younger than 50 years, and Peterson et al. 34 , studying other co-morbidities (cancer, diabetes mellitus, systemic arterial hypertension), found no association with inappropriate ED use.
In relation to self-perceived health, three authors 12,13,34 found no association with the outcome.However, Sarver et al. 25 pointed out that individuals with worse self-perceived health had nearly three times more inappropriate ED visits than those with good to excellent self-perceived health, but in this case self-perceived health was assessed outside the ED.

Access to health services • Cost and form of payment for care
According to three authors, cost of the consultation was associated with inappropriate use.Oktay et al. 21found that those who paid for the consultation used the ED 60% more appropriately than those who did not pay.Lee et al. 32 attributed the preference for using primary care (rather than the ED) for outpatient complaints to the lower cost of primary care, while Young et al. 35 ascribed higher inappropriate ED use to the fact that it cost less.Reasons for ED rather than primary care: ease of access (32%), trust (7%) familiarity (11%), and no reason (7%).
Multivariate:  After adjusting for possible confounding factors, four studies 13,22,25,34 found no association between not having a health plan and inappropriate ED use, two of which 13,22 were conducted in countries with public health care systems and two in countries without such a system 25,34 .reira.
et al. 22 Bianco et al.No association with "kind of insurance" (p = 0.465) or day of week (p = 0.814).• Regular physician and regular health care facility

Multivariate
In two studies, individuals with a regular physician showed 40 and 67% less inappropriate ED use, respectively 26,34 .In the same sense, two other studies showed that individuals with a link to primary health care used the ED more appropriately 31,35 .However, four studies found no such associations for regular physician 13,32 or regular source of care 12,13,26 .This lack of association may be related to the more rigorous criteria for defining inappropriate use 12,13 and the low precision of the multivariate analysis, suggested by wide confidence intervals found in one of the studies 32 .nn et al. 40 Rodríguez et al. 23 Stein et al. 26

Referral
According to Coleman et al. 44 , two-thirds of patients with inappropriate complaints came directly from home.Five studies showed that individuals who went to the ED on their own had 1.39 to 2.42 times greater odds of using the ED inappropriately, as compared to those referred by a health professional 13,14,35,37 or other hospitals 16 .Meanwhile, two authors 26,45 did not find this association, but Barker et al. 45 observed that patients who had been seen previously by a physician but not referred showed more inappropriate use.

• Prior consultation for the current complaint
The findings related to having a prior outpatient consultation are contradictory.An Australian study 40 found that individuals in an aged care facility who had prior contact with a physician for their current problem consulted the ED more appropriately (p = 0.02).Meanwhile, Dale et al. 19 found that patients who had prior contact with a health professional for their current complaint had more inappropriate ED visits (p < 0.001).Carret et al. 13 , in a multivariate analysis, showed no such association.

• Access
The four descriptive studies that evaluated the access-related reasons for preferring the ED detected difficulties in consulting primary care, such as the primary care facility being closed, difficulty in obtaining an appointment, and waiting time as reasons for inappropriate ED use 12,13,16,21,32,35,43 .Lee et al. 32 and Rajpar et al. 43 found that 35% and 50% of patients, respectively, reported closed primary care facilities as one of the reasons for inappropriate ED use, but in the latter 46.3% had not even attempted contact with primary care.Two analytical studies found an association between difficult access to primary health care and inappropriate ED use.In the cohort study in the United States, this association had a p of 0.029.The variables comprising this indicator, difficulty in scheduling primary care, difficult telephone contact for primary care, and longer waiting time for a primary care appointment were also associated with inappropriate ED use (p < 0.03) when evaluated singly in the multivariate analysis 25 .Meanwhile, in the Brazilian study, difficulty in obtaining a primary care appointment, refusal by the primary care physician to Cad.Saúde Pública, Rio de Janeiro, 25(1):7-28, jan, 2009 treat patients without a previously scheduled appointment, and primary care being open for shorter hours were associated with inappropriate ED use in the 15-49-year bracket (RR = 1.38, 1.44, and 1.63, respectively) 13 .

• Characteristics of ED use
Ease in obtaining diagnostic tests and treatments at the ED were other reasons cited for preferring the ED 32,33,35,43 .Rajpar et al. 43 and Lee et al. 32 reported that 3.7% and 11.4%, respectively, preferred the ED due to greater ease of diagnosis.Young et al. 35 reported that 21% of patients chose the ED claiming better care or greater ease of diagnosis and treatment.Carret et al. 13 , in a multivariate analysis, found no association between difficulty in obtaining tests, medicines, and referral to specialists through primary care and inappropriate ED use.
Sempere-Silva et al. 16 pointed to trust in the service provided by the ED as the principal reason for inappropriate use of this service, while in Afilalo et al. 12 , 7.1% of patients cited this reason.Sarver et al. 25 , in a multivariate analysis, found that the greater the dissatisfaction with primary care, the greater the inappropriate use of the ED (RR = 1.13 per unit, in a score ranging from 0 to 4; 95%CI: 1.01-1.25).

Principal diagnosis
The principal diagnoses found in persons that visited the ED inappropriately varied considerably from study to study, depending on the classification used in the target age bracket.Three descriptive studies indicated that the principal diagnosis or complaints among persons that consulted the ED inappropriately were respiratory problems (especially in younger patients) and digestive problems 42 ; abdominal pain and chest pain 12 , and eye and ear complaints 16 .

Discussion
The prevalence and factors associated with inappropriate ED use varied widely, depending mainly on the criteria used and the study population.The criteria often underestimate inappropriate ED use.This occurs when using self-perceived urgency as the criterion 6,37 or when the criteria define as appropriate all cases requiring intravenous medication or any type of complementary test 13,14,16,36,41 .In this situation, the definition ends up including problems that require immediate attention, but which occur when the primary health care facility is open and operating and could thus be resolved at this level of care (as in the frequent case of patients with fever, high blood pressure, dysuria, and other complaints).
This literature review indicates that despite the above caveats, the prevalence of inappropriate ED use was consistent in a large number of studies, even across countries with different health care systems.The studies point to a high prevalence (from 20 to 40%), concentrated more in daytime ED visits.The principal factors associated with inappropriate ED use were younger age, female gender, absence of co-morbidities, lower health spending, not being referred by a health professional, not having a regular physician or regular source of care, and difficulty in accessing primary care.No association was found between marital status, occupational status, or self-perceived health and inappropriate ED use.The associations with other variables such as race or skin color and prior consultation for the current complaint showed varied and sometimes contradictory results.
The associations between younger age and female gender and inappropriate ED use were highly consistent 13,14,21,22,25,30,34,37 .Among the elderly, the prevalence of chronic non-communicable diseases is high.In this group, ED use is considered appropriate in most cases, even in situations that could be managed at more basic levels of care, since even in such cases the patients end up having complementary tests or receiving IV medication 9,46 .As for female gender, the association could be explained by the fact that women use health services more in general and maintain this same trend with inappropriate ED use 47 .
Cad. Saúde Pública, Rio de Janeiro, 25(1):7-28, jan, 2009 The studies point to a direct association between socioeconomic status and inappropriate ED use 21,22,31 .This finding is plausible, since higher-income individuals have greater access to health services in general, suggesting health inequity 47 .This ends up aggravating health inequity since higher-income individuals occupy the ED with inappropriate complaints, thereby limiting access by the lower-income population that not only has greater health needs but also comes to the ED more often for truly urgent problems.Inequity also reduces the emergency department's efficiency, since patients that most need ED services are the ones with the least access.The lack of association with socioeconomic status in some studies might be related to: methodological specificities, especially in data analysis with only two categories or with a high cutoff point for the schooling variable, thus hindering the identification of associations or impeding evaluation of tendencies 34 ; the use of only univariate analysis 12 ; inclusion of all the variables at the same time in the analytical model, causing adjustment for mediating variables; or colinearity between income and schooling and underestimating the effect 48 .
The negative association between presence of co-morbidities and inappropriate ED use appears to be related to greater health needs in these cases, which require more detailed examination by the physician.Thus, most cases of co-morbidities ended up being classified as urgent 9 .In addition, as Petersen et al. 34 showed, the association between chronic disease and inappropriate ED use appears to vary according to the type of chronic disease.Carret et al. 13 only found a negative association between presence of chronic non-communicable diseases and inappropriate ED use in the elderly.In younger patients this association was not found, perhaps due to lack of statistical power, either because chronic diseases are less prevalent in the younger age brackets or because the elderly present chronic diseases of a different type or severity.This age effect modification may explain the fact that some authors did not find an association when they studied populations that included young adults 13,37 .
The majority of the authors did not find an association between self-perceived health and inappropriate ED use.The only author that found an association between poor self-perceived health and inappropriate ED use collected this variable in a household setting 25 .These findings indicate that measurement of self-perceived health in the ED is heavily affected by the moment in which this question is asked, and that it is difficult to interpret.This measurement of selfperceived health does not appear to reflect the way individuals assess their overall health status, an aspect that relates more to the way health services are used.
In relation to factors associated with access to the ED, it was clear that the cost concern influences the decision about the place of consultation.This is reinforced by studies we reviewed showing that individuals seek levels of care with the lowest cost and consult the ED less inappropriately when they have to pay 31,32,35 .Some studies that evaluated the relationship between health plan coverage (yes versus no) and inappropriate ED use did not find such an association 13,22,25,34 , while according to Liu et al. 30 the association can vary according to the type of health plan.The findings suggest that individuals with high income and excellent health plans rarely consult the ED, that they have easy access to their regular physicians and services even outside of business hours, and that if necessary they can even be admitted to hospital without going through the ED.Meanwhile, individuals with health plans that impose a wide range of limitations on both coverage and speed of access appear to be similar to users of the public health system in terms of inappropriate ED use.
The same logic applies to the regular physician and regular source of care.To avoid inappropriate use, it appears not to be sufficient to have a fixed physician and regular source of care; rather, this physician or service must be easily and quickly accessible 49 .In addition, the regular doctor and regular source of care appear to have a greater impact on reducing inappropriate ED use in cases involving doubt about urgency.This is reinforced by the fact that studies showing an association between regular physician/ regular source of care and inappropriate ED use used a less rigorous criterion for defining such use 31,34,35 as compared to studies that found no such association 12,13 .The importance of having access to physicians at other levels of care is also demonstrated by the consistency of the positive association between referral by a health professional and appropriate ED use 13,14,35,37 .
Another aspect evaluated was prior consultation for the current complaint, which did not show consistent findings.One reason for this inconsistency was the fact that the studies evaluated different populations, such as residents of homes for the elderly, who may often even need an authorization from the institution's physician to be referred to the ED 40 .Another relevant aspect is the imprecision in defining exposure, i.e., failure to specify the time during which the prior consultation for the current complaint occurred, and the difficulty in demarcating the current complaint, thus hindering interpretation of the findings.Most of the studies on reasons for preferring the ED are descriptive 16,21,35,43 .The small number of studies of association on reasons for preferring the ED help explain the scarcity of accumulated knowledge on the subject.The existing studies suggest that difficulty in access to primary health care is an important factor for inappropriate ED use 12,13,16,21,32,35,43 , although the difficulty is not simply the availability of primary care services (because the prevalence of inappropriate ED use increases during the day, when primary care is operating), but the possibility of obtaining care quickly 3,49 .Patients think it is so unlikely that they will get a timely primary care appointment that many do not even try 49 .They fact that they receive tests and treatment during the ED visit and their trust in the services provided by the ED are important enticements for choosing the ED 32,35,43 .However, typical difficulties in primary health care such as delays in performing tests and obtaining medicines and referrals to specialists were not associated with inappropriate ED use 13 .Rather, this outcome was associated with dissatisfaction over primary care 25 .
Inappropriate ED use occurs in various parts of the world, thereby increasing costs and hindering access to the ED for really urgent cases 3,50 .In addition, non-urgent cases are managed hastily and fail to benefit from the comprehensive care and continuity they could obtain at other levels of care 49 .
There are various studies on inappropriate ED use, but the diversity of criteria used limits the comparability of findings.Nevertheless, the socio-demographic profile of the population that consults the ED inappropriately is well defined.Meanwhile, most available studies on the reasons for consulting the ED inappropriately are descriptive, so these aspects have not been fully clarified.
This review indicates that access to primary health care plays a key role in reducing inappropriate ED use.However, intervention studies have already demonstrated that expanding the amount of services, business hours, and health professionals in primary care does not have a major impact in reducing inappropriate ED use 10,51 .This study suggests that primary care needs to provide high-quality reception of patients, with efficient triage so as to treat cases quickly when they cannot wait 4,49 .In countries with a public health care system, like Brazil, this can be facilitated by continuity in health care, since the primary care team establishes a bond and takes responsibility for the population in its coverage areas and is thus better prepared to provide adequate reception and efficient triage.In parallel, it is important to inform the population on situations in which they really should use the ED and the disadvantages of consulting the ED when the case is not truly urgent 5 .The emergency department, in turn, should develop a counter-referral system to make the tests, diagnoses, and prescriptions performed by the ED available to the other levels of care, in order to improve the care in general and avoid repeating unnecessary tests.Other measures are needed with specific populations, like those that consult the ED repeatedly ("frequent flyers"), aimed at redirecting them to other levels of care, thereby helping minimize the problem.
Future studies, especially on association, with standardized criteria to characterize inappropriate ED use, with in-depth investigation of the reasons leading the population to use emergency services inappropriately, can help establish strategies to reduce the problem.Intervention studies will also be needed to test the proposed strategies.In-depth development is also needed for criteria to define inappropriate use in order to better discriminate between cases that may need immediate attention but can be managed by primary care and those that not only require immediate attention but also require procedures not available in primary care facilities.
and non-urgent: levels: emergency, urgent, and non-urgent.

Table 1
Criteria for defi nition of inappropriate emergency department use, 2007.

Table 2
Characteristics of studies that evaluated factors associated with inappropriate use of the emergency department (ED).