Emergency Department Use among Patients with Mental Health Problems: Profiles, Correlates, and Outcomes

Patients with mental health (MH) problems are known to use emergency departments (EDs) frequently. This study identified profiles of ED users and associated these profiles with patient characteristics and outpatient service use, and with subsequent adverse outcomes. A 5-year cohort of 11,682 ED users was investigated (2012–2017), using Quebec (Canada) administrative databases. ED user profiles were identified through latent class analysis, and multinomial logistic regression used to associate patients’ characteristics and their outpatient service use. Cox regressions were conducted to assess adverse outcomes 12 months after the last ED use. Four ED user profiles were identified: “Patients mostly using EDs for accessing MH services” (Profile 1, incident MDs); “Repeat ED users” (Profile 2); “High ED users” (Profile 3); “Very high and recurrent high ED users” (Profile 4). Profile 4 and 3 patients exhibited the highest ED use along with severe conditions yet received the most outpatient care. The risk of hospitalization and death was higher in these profiles. Their frequent ED use and adverse outcomes might stem from unmet needs and suboptimal care. Assertive community treatments and intensive case management could be recommended for Profiles 4 and 3, and more extensive team-based GP care for Profiles 2 and 1.


Introduction
Patients with mental health (MH) problems, including mental disorders (MDs), substance-related disorders (SRDs), and suicidal behaviors, are known to use emergency departments (EDs) frequently.A 2016 systematic review and meta-analysis [1] and a 2021 US study [2], respectively, found that 4% and 5.3% of ED visits were for MDs, including SRDs.A Canadian study found that in 2014-2015, 17% of ED users with MDs or SRDs had visited the ED at least four times for any reason, and that 22% of these patients had visited the ED on three to five consecutive years [3,4].Some patients may thus be "recurrent" ED users who use EDs repeatedly over several years; some of them will be high or very high ED users, respectively, defined as 3+ [5,6] or 8+ [7,8] ED visits/year.Patients may use the ED for severe conditions such as suicidal behaviors, or for milder ones treatable in outpatient care [9].Some use the ED in a more seasonal manner, during periods where they may be having a harder time-e.g., the holiday season [10].The ED may also be the entry point to care for patients with incident MDs (new cases), which according to some studies could represent as many as 45% of all patients using the ED [3,11].While EDs provide rapid access to acute MH care [12], limited access to outpatient care or receiving inadequate care may also lead to frequent ED use [13], thus contributing to ED overcrowding [14], higher healthcare costs [15], and adverse outcomes (e.g., hospitalization, death) [16].ED users with MH problems are a heterogeneous population.Using person-centered approaches like latent class analysis, which correlates specific user characteristics rather than variables on heterogeneous populations, may present solutions for finding distinct profiles among ED users [17].Improving our understanding of ED user profiles by associating them with patient characteristics, service use and subsequent adverse outcomes could help recommend care that is better adapted to these patients' needs.
Several studies assess profiles of patients who use the ED frequently for any medical reasons, identifying four [18][19][20][21][22] or five [23,24] high ED user profiles, with one [20,22] to four [23,24] profiles mainly comprising patients with MDs or SRDs.None of these studies link ED user profiles to subsequent health outcomes.Few studies identify ED user profiles exclusively among patients with MDs [25 -27].One study found four profiles; the profile with the largest number of high ED users also reported the lowest quality of life [25].Another study on patients with incident MDs identified five profiles: the profile with the most complex and serious health conditions included more high ED users, had the lowest continuity of care from general practitioners (GPs) and psychiatrists, and the worst adverse outcomes [26].Another study investigating high ED users with MDs identified three profiles mainly made up of 3-year recurrent very high ED users and 2-year recurrent high ED users with poor health conditions and high risk of death [27].Recurrent high and very high ED users were reported to be mostly men [21] of lower socioeconomic status [28] with worse health and social conditions; most had a severe MD [27], were high users of health services in general, but did not necessarily receive adequate care [27].These characteristics are usually associated with a higher risk of hospitalization and premature death [29,30].
To our knowledge, no prior study has assessed ED user profiles by considering patients with various MH problems (MDs, SRDs, and suicidal behaviors), whether incident cases or not, and by investigating the intensity, recurrence, triage priority, and dispersion of ED use over a 4-year period.This study is original in that it also examined outpatient service use over the 12 months before the last ED use to assess whether more continuous, intensive, diversified, and regular outpatient service use could be associated with ED user profiles and subsequent risks of hospitalization and premature death.Most previous studies have used hospital administrative records instead of provincial databases that would include broader information on outpatient care use.Such information could contribute to our understanding of ED use profiles.Based on provincial databases, this longitudinal study thus aimed to identify profiles of ED users with MH problems, to associate these profiles with patient sociodemographic and clinical characteristics and service use, and to subsequently link them to adverse outcomes.

Study Design, Cohort, and Data Sources
This study investigated a 2012-2017 cohort of 11,938 patients with MH problems who had used one of six psychiatric EDs in Quebec (Canada).To be eligible, patients had to be enrolled in Quebec Health Insurance Plan, aged 12 or older, and have used an ED between 1 April 2012 and 31 March 2016.Patients hospitalized for over 120 days in the year before their last ED visit were excluded, to ensure accurate assessment of outpatient care.Patients were identified by their name and their medical insurance number from the Quebec Health Insurance Plan.The Quebec Health Insurance Plan databases used for this study included the health insurance registry (FIPA, containing patient sociodemographic information, i.e., sex, birth date, and age), the physician claims database (RAMQ), the ED use database (BDCU), the hospital inpatient database (MED-ECHO), the community healthcare center database (I-CLSC, mostly psychosocial services), and the vital statistics death database (RED).Variables that appeared in several databases were merged (see Table 1 footnotes).Diagnostic codes were classified according to the International Classification of Diseases, ninth or tenth revisions (Appendix A).This study followed the Strobe guideline for epidemiological studies [31].The ethics committee of the Douglas Mental Health University Institute approved the study protocol (IUSMD-20-26).

Study Variables
The variables used to formulate patient profiles included: ED use for incident MDs or using the ED as a primary MH service (measured from 1 April 2014 to 31 March 2015); number of years of ED use; mean number of ED visits/year; very high ED use; recurrent high ED use; percentage of ED visits with high triage priority; low ED use dispersion.All these were monitored from 1 April 2012 to 31 March 2016.As in previous research, incident MDs were defined as diagnoses made after a 2-year period without any MD, SRD, or MH consultation [11].Very high ED use was defined as 8+ visits/year [7], high ED use as 3+ visits/year [5], and recurrent high ED use meant these visits occurred over 2+ consecutive years [32]-all minimum benchmarks.In agreement with the Canadian Triage Acuity Scale, ED triage priority assessed illness acuity on a scale of 1 to 5, with level 1 representing the most urgent cases and levels 4 and 5 deemed suitable for outpatient care [33].The proportion of high-priority visits (levels 1-3) was calculated out of all ED visits (levels 1-5).Dispersion was used to measure ED use over the year, with patterns that could occur over a short period, such as during the holidays [21].Having 2+ ED visits within a 90-day period over 2+ consecutive years was defined as "low dispersion" [34].
Sociodemographic correlates included sex at birth, age group, residing in more materially or socially deprived areas, all recorded at the initial ED use; violent/disturbed behaviors or social problems were documented from the first to the last ED use.Based on the smallest geographic areas outlined in recent versions of the Canadian census, Material and Social Deprivation Indexes were categorized into two groups: the least and moderately deprived areas (quintiles 1-3); the more deprived areas (quintiles 4-5, including areas not assigned [e.g., homelessness]).The Material Deprivation Index gauged employment rates, average income, and the proportion of residents without a high school diploma, while the Social Deprivation Index assessed the proportion of residents living alone, without a spouse, and single-parent families [35].Violent/disturbed behaviors and social problems (e.g., homelessness) were identified by ED triage nurses.
Clinical correlates were assessed between the first and last ED use and included the principal MD, SRDs (alcohol/drug use, induced disorders, intoxication, withdrawal), suicidal behaviors (suicide ideation/attempt), chronic physical illnesses (e.g., cardiovascular illnesses) and their severity, and ambulance/stretcher use.The principal MD were categorized as a serious MD (schizophrenia spectrum and other psychotic disorders, bipolar disorders), personality disorders, or a common MD (depression, anxiety, adjustment, attention deficit/hyperactivity disorders).Chronic physical illnesses were assigned a severity score ranging from 0 to 3 adapted from the Elixhauser and Charlson Comorbidity Indexes, with higher scores indicating a higher risk of death [36].The Charlson Comorbidity Index measures 17 medical conditions [37]; the Elixhauser measures 30.Together, they cover 32 chronic physical illnesses [36].Being transported to an ED by ambulance or using a stretcher served as proxies for the severity of conditions, calculated by dividing the number of times these measures were utilized by the patient's total number of ED visits.
Measured in the year prior to the last ED visit, service use correlates included having a usual GP or psychiatrist, high continuity of physician care, frequency of MH outpatient consultations by any physician, psychosocial interventions from community healthcare centers, and high regularity of outpatient care.A proxy for family physician, "usual GP" required a minimum of two consultations with the same GP or with a GP in a family medicine group, which is a medical clinic where GPs work in groups that include clinicians, mostly nurses and social workers; patients are also registered in these clinics, which offer comprehensive patient care [38]."Usual psychiatrist" required 2+ outpatient consultations with the same psychiatrist, or one psychiatric consultation showing collaborative care with the usual GP [39].Continuity of physician care was gauged using the Usual Provider Continuity Index [40], which calculates the proportion of consultations with the usual physicians relative to all physicians consulted-a score of ≥0.80 indicates high continuity of care [41].The frequency of outpatient consultations measured care intensity, 4+ consultations/year being the minimum benchmark [42,43].In Quebec, public psychosocial services are mainly provided by community healthcare centers [44].High regularity of outpatient care integrated care received from usual physicians and psychosocial services, requiring at least one patient-clinician contact in each of the four 3-month periods of the year, which indicated close patient follow-up care over time [45].
Adverse outcomes were monitored for 12 months after the last ED use (2012 to 2016) and included hospitalization for MH reasons and death for any cause.These metrics are key indicators for measuring adverse outcomes utilizing health administrative databases [46][47][48].

Data Analysis
Considering that missing values were less than 1%, a complete case analysis was used [49].Univariate analyses included frequency distributions for categorical variables.Latent class analysis (LCA) [50,51] was used to identify patient profiles based on ED use variables.Compared to standard cluster analysis, LCA provides a stronger alternative for assessing model fit and captures classification uncertainty more effectively [52].A series of increasingly complex models (adding classes) was evaluated to determine the optimal number of latent classes.Following standard practice, the Bayesian information criterion (BIC) [53] and the entropy value [54] were calculated to identify the best model.Comparative analyses of correlates were conducted across patient profiles using bivariate multinomial logistic regression, based on the patient sociodemographic and clinical characteristics, and service use.Relationships between patient profiles and adverse outcomes were tested using Cox regressions, adjusted for age and sex.Unadjusted risk ratios and adjusted hazard ratios were calculated.LCA was performed with SAS 9.4 [55] and other analyses were carried out with Stata 17 [56].

Cohort Description
Of the initial 11,938 patients, 256 were excluded because they had been hospitalized >120 days in the 12 months before their last ED use.In the final cohort of 11,682 patients, 24% had incident MDs, 20% had used EDs over the 4-year period, 25% had 4+ ED visits/year, 13% were very high ED users, 30% had high triage priority in 67-100% of ED visits, and 37% showed low ED use dispersion (Table 1).Half the patients (51%) were women; 55% were 30-64 years old and also lived in more materially or socially deprived areas.About half had a serious MD, 40% had SRDs, and 44% exhibited suicidal behaviors or chronic physical illnesses.In the 12 months before the last ED use, 50% had a usual GP, 39% a usual psychiatrist, 61% showed high continuity of physician care, 40% and 53% did not receive care from any outpatient physician or community healthcare center, respectively, and 37% received a high regularity of outpatient care.In the 12 months following the last ED use, 10% were hospitalized and 2% died.

Patient Profiles
The four-class solution was selected based on the smallest BIC (three-class = 3293; four-class = 2361; five-class = 2368), with an entropy value of 0.84->0.8indicating a limited class overlap.Accounting for 30% of the sample, Profile 1 included the most patients with incident MDs (47%) or who use an ED as their first MH service (Table 2).Most Profile 1 patients (83%) had a mean of one ED visit/year, with 44% using an ED only once in the four years considered, 27% using it over 2 years, and 25% over 3 years.Profile 1 included the most patients (45%) with high triage priority on 67-100% of ED visits.None were recurrent high or very high ED users.Profile 1 was labelled "Patients mostly using EDs for accessing MH services".Profile 2 (32% of sample) included the most patients with 2 years of ED use and a mean of 2 ED visits/year.It had the least number of patients with high triage priority on 67-100% of ED visits and included no recurrent high ED users-though 4% were very high ED users.Profile 2 was labelled "Repeat ED users".
Profile 3 (19% of sample) included the most patients with 3 years of ED use and a mean of 3 ED visits/year.In this profile, 29% were recurrent high ED users, but none were very high ED users.Almost all Profile 3 patients had 2+ visits within 90 days over at least 2 years and showed low ED use dispersion.Profile 3 was labelled "High ED users".
Profile 4 (19% of sample) comprised the most patients with 4 years of ED use.Almost all of them averaged 4+ ED visits/year, with low ED use dispersion.Most (82%) were recurrent high ED users, and 61% very high ED users.Profile 4 was labelled: "Very high and recurrent high ED users".

Patient Profiles and Associated Sociodemographic and Clinical Characteristics and Service Use
Compared to Profile 1, Profiles 4, 3, and 2 patients were 22%, 18%, and 12% more likely to be women, and 83%, 70%, and 21% more likely to be 65+ years old, respectively.Profiles 4 and 3 patients were also 42% and 25% more likely to be 30-64 years old compared to those in Profile 1. Profiles 4, 3, and 2 patients were 89%, 47%, and 29% more likely to live in materially and socially deprived areas and had five or three times and 88% higher risk of exhibiting violent/disturbed behaviors or social problems, respectively, than those of Profile 1.
Compared to Profile 1, Profiles 4, 3, and 2 had 30, 9, and 3 times more risk of having a serious MD as principal MD; 12, 5, and 2 times more risk of personality disorders; 3 or 1 times and 85% to have a common MD; and 4 or 2 times and 36% SRDs, respectively.Compared to Profile 1, the risk of suicidal behaviors for Profiles 4, 3, and 2 was 1 time, 40% and 14% higher, and the likelihood of chronic physical illnesses with a 3+ severity score being 2 or 1 times and 14% higher, respectively.And compared to Profile 1, Profiles 4, 3, and 2 patients had an 81%, 80%, and 38% lower risk of being transported to EDs by ambulance or on a stretcher at least 50% of the time, respectively.
Profiles 2, 3, and 4 patients were 28%, 17%, and 14% more likely to have a usual GP, and were 1, 2, and 4 times more likely to have a usual psychiatrist, respectively.Profiles 2 and 3 patients were 17% and 16% more likely to have a high continuity of physician care compared to those of Profile 1. Profiles 4, 3, and 2 patients had 94%, 71%, and 58% more chances of receiving at least four MH outpatient consultations from any physician, and were four and two times, and 73% more likely to receive at least four interventions from community health centers, respectively, than those of Profile 1. Profiles 4 and 3 patients were also four and two times more likely than those in Profile 1 to receive a high regularity of outpatient care (Table 3).

Associations between Patient Profiles and Adverse Outcomes
Compared to Profile 1, risks of hospitalization for MH reasons were five times higher in Profile 4, three times higher in Profile 3, and 74% higher in Profile 2. Risk of death for any cause was higher in Profiles 4 and 3 (96%, 66%) than in Profile 1 (Table 4).

Discussion
This study examined patients with MH problems over a 4-year period in order to identify ED user profiles, link correlates to these profiles, and assess subsequent risk of hospitalization and death.As previous ED studies have shown [57,58], patients in this study were quite vulnerable overall: about half were poor or isolated, affected by serious MDs, SRDs, chronic physical illnesses, or suicidal behaviors.One quarter of patients had a mean of 4+ ED visits in the 4-year ED follow-up period.This number is higher than that found in previous studies [59,60], which could be because patients were only recruited in large psychiatric EDs and had conditions such as suicidal behaviors.About half did not receive any physician or psychosocial service in the year before their last ED use, and only a third received a high regularity of care.The percentage of first ED use for incident MDs found here (24%) was lower than in other studies (±50%) [3,11], which could also be due to patients only assessed in psychiatric EDs.Four profiles of ED users with MH problems were identified, each substantially different from the others.
Accounting for one-third of the cohort, Profile 1 patients differed from other profiles as they used EDs the least during the 4 years considered.About half of them had entered the MH system through EDs and had not received MD or MH care during the 2 previous years; 83% showed a mean of only one ED visit.Roughly half showed only 1 year of ED use, about a quarter had 2 years, and the other quarter, 3 years.Profile 1 ED use was deemed the most urgent by the triage nurses, probably because they lacked MH care and had waited to have acute symptoms before going to the ED.The fact this profile included more men and younger patients may also explain these care-seeking behaviors: compared to women and older patients, men and younger patients are known to use less MH services, and only as their last resort [61,62].Profile 1 patients also showed lower risk factors: they were not as poor and isolated and had fewer complex conditions than Profiles 4 and 3, especially in terms of serious MDs and SRDs.These characteristics explain why, of all profiles, Profile 1 patients also received the least MH care within the year of their last ED use.Fewer of them received care from psychiatrists and psychosocial services, and they had the lowest regularity of care.Their low ED use and MH conditions explain why they showed the lowest risks of hospitalization and death.Profile 1 patients may be easily treated in primary care [63], may benefit from extended team-based GP care (nurses, social workers) mirroring the chronic care model [64], including collaborative care with psychiatrists [65,66].Brief intervention ED teams (e.g., ED liaison agents, ED case management) [67,68] might also be beneficial as they would provide care to these patients before they could be transferred to adequate outpatient care.
Accounting for about one-fifth of the cohort, Profile 4 contrasted from Profile 1 as its patients had the worst ED use, poor social and material conditions, violent/disturbed behaviors, and were most affected by serious MDs, SRDs, suicidal behaviors, and severe chronic physical illnesses, all of which are characteristics of very high and recurrent high ED use [27,58,69,70].The fact this profile featured very high ED users may explain why about half of them showed high triage priority on 33 to 66% of ED visits.They received the least GP care after Profile 1, but even though they were high ED users, they received the most psychiatrist and psychosocial care-high ED users are known to also be high MH care users [71,72].Patients with these conditions are usually more likely to be followed in psychiatry rather than primary care, since GPs are often uncomfortable treating patients with these severe conditions [73,74].Profile 4 patients received more outpatient care but may still have elevated unmet needs, as their continuity of care did not differ from that of Profile 1.These characteristics explain why Profile 4 showed the highest risk of hospitalization and death-hospitalizations are known to be frequent among patients affected by serious MDs and chronic physical illnesses [75].Perhaps Profile 4 patients also had a higher risk of death because they were older and had severe chronic physical illnesses and suicidal behaviors-the latter being also associated with MDs and SRDs [76].These patients may be flagged in ED medical records, so information about their high ED use can be made readily available to ED clinical teams to make sure they receive appropriate intensive care and have an individual care plan.ED staff could also work closely with outpatient clinicians to ensure these patients receive adequate care for their serious conditions.Assertive community treatments integrating SRD treatments [77] might also be suggested for very high and recurrent high ED users such as those in Profile 4.
Encompassing almost 20% of the cohort, Profile 3 resembled Profile 4, but with milder health-seeking behaviors and risk factors.These patients were mostly high ED users, with about a third using EDs recurrently over 3 consecutive years.Like those of Profile 4, they were regular ED users, with 2+ ED uses in each 3-month period over 2+ years.Compared to other profiles, they were transported more often to EDs by ambulance and/or on a stretcher.It is possible that Profile 3 included more involuntary ED visits than Profiles 1 and 2, as patients with serious MDs and co-occurring disorders, violent/disturbed behaviors or social problems, and who are isolated and older are more prone to involuntarily solicit acute care [78].Risks of hospitalization and death were lower in Profile 3 than in Profile 4, but higher than in the two other profiles.Intensive case management [79,80] and integrated MD-SRD treatments [77] may be recommended for Profile 3 patients.Intervention plans [81] may also be developed in the ED in close collaboration with outpatient teams to better target these patients and reduce their high ED use.
Accounting for 32% of the cohort, Profile 2 patients were characterized by repeated ED use, nearly half of them having 2 years of ED use.High triage priority was the lowest of all profiles: only 20% showed high priority 67-100% of the time.This means that, in most cases, these patients could have avoided using the ED.Profiles 2 and 1 had almost identical percentages in each age group and showed similar regularity of outpatient care, though Profile 2 patients had worse conditions and received more outpatient care than Profile 1-but less than Profiles 4 and 3.More Profile 2 patients had a usual GP than those of Profiles 3 and 4, perhaps because they had less serious MDs and co-occurring disorders [82].These conditions give Profile 2 a higher risk of hospitalization compared to Profile 1, though risk of death was similar in both.Like those of Profile 1, Profile 2 patients may benefit from extended team-based GP care coupled with more psychosocial services, including help for crisis resolution [83] and peer support, as these may prevent the types of ED use nurses often perceive as non-urgent.Access to brief intervention teams providing follow-up care after ED use [68,84], transitioning patients to outpatient care, short-stay crisis units [85], and home-based treatments [86] might also be appropriate for these patients to avoid acute care.
This study has limitations.First, the study database did not include information on issues linked to ED use like psychotic crises or high levels of stress, on programs such as assertive community treatment or intensive case management, on services received from private sector psychologists and addiction treatment centers, or on community-based services (e.g., crisis centers).Information on these services would have improved our understanding of outpatient care use and of its impact on ED use.However, even as these services are key in responding to patient needs, they remain mostly underfunded in Quebec and generally offer limited services to patients like those in this cohort [44,87,88].Second, this study only measured hospitalization and death as adverse outcomes because other outcomes are unfortunately not documented in health administrative databases.Third, patients with long periods of hospitalization before their last ED use were excluded from this study, but they only comprised 2% of the sample.Fourth, materially or socially deprived conditions were based solely on the area where the patient lived.Some patients may live in poor areas but show a high household income, others may live alone or be single parents and have a strong social network, but unfortunately the study database did not contain information on the patients' household income or social network.Fifth, it was hard to compare the profiles identified in this study with those found in the previous literature, as research is scarce in this area, and rarely considers similar variables.Finally, the findings may not be generalizable beyond psychiatric EDs, large urban settings, or contexts lacking universal health coverage for vulnerable populations.

Conclusions
This study identified four ED use profiles among patients with MH problems.About a quarter of patients showed a mean of 4+ ED visits in the four consecutive years that were considered.These patients were quite vulnerable and needed more adequate care to respond to their substantial needs.Profiles 4 and 3 had the most ED use, the most severe conditions and received the most outpatient care, demonstrating the fairness in accessibility of Quebec's public health system, which prioritizes the most vulnerable populations.The risk of hospitalization and death was also higher in these profiles, especially when compared to Profile 1 patients, who used EDs less frequently and primarily to access MH care.The frequent ED use and adverse outcomes of Profiles 4 and 3 might stem from unmet needs and suboptimal care.Based on this study's findings, more intensive interventions are recommended for Profiles 4 and 3, including integrated MD-SRD treatments, assertive community treatments, and intensive case management.Profiles 2 and 1 could benefit from extended team-based GP care and crisis resolution teams.Strategies like intervention plans, short-stay crisis units, and home-based treatments should be more widely deployed in EDs to reduce frequent and recurrent ED use.Overall, patients should be more accurately identified in the ED, and integrated work should be carried out with outpatient services.Future research should favor mixed methods to investigate the types of outpatient care that are not considered in administrative databases and evaluate their impact on ED use.Understanding the perceived unmet needs that can lead to ED use may also be important to improve patient care and reduce ED use.F111, F131, F141, F151, F161, F181, F191, F112, F132, F142, F152, F162, F182, F192 (drug abuse or dependence); F113-F114, F133-F134, F143-F144, F153-F154, F163-F164, F183-F184, F193-F194 (drug withdrawal) F115-F119, F135-F139, F145-F149, F155-F159, F165-F169, F185-F189, F195-F199 (drug-induced disorders); F110, F130, F140, F150, F160, F180, F190, T400-T406, T408, T409, T423, T424, T426, T427, T435, T436, T438, T439, T509, T528, T529 (drug intoxication) D500; K257, K259, K267, K269, K277, K279, K287, K289; B20-B24; D65-D68, D691, D693-D696; B18, I85, I864, I982, K700-K703, K709 K711, K713-K715, K716, K717, K721, K729, K73, K74, K754, K760, K761, K763, K764, K765, K766, K768, K769, Z944; L900, L940, L941, L943, M05, M06, M08, M120, M123, M30, M31, M32-M35, M45, M460, M461, M468, M469; G041, G114, G80, G81, G82, G83; E40-E46, R634, R64, D51-D53, D63, D649; D501, D508; D509 a All diagnoses identified in RAMQ (Régie de l'assurance maladie du Québec, Physician Claims Database) for the full study period were based on the International Classification of Diseases Ninth Revision (ICD-9), which included a 4-digit code, for the financial year 1 April to 31 March.The Canadian Tenth Revision (ICD-10-CA) was used in MED-ECHO (Maintenance et exploitation des données pour l 'étude de la clientele hospitalières, Hospital Inpatient and Day Surgery Database) and in the BDCU (Banque de données communes des urgences, Emergency Department Use Database).All diagnoses related to the above databases were considered, and all data integrated each year, for each patient.MED-ECHO is the only database that includes several diagnoses: principal diagnosis and numerous secondary diagnoses.In the databases used in this study, MDs were considered only as principal diagnoses, but SRDs as both principal and secondary diagnoses, considering that SRDs are often underdiagnosed.b Suicide attempts were identified in MED-ECHO (principal and secondary diagnoses), and suicidal behaviors (suicide ideation/attempt) detected in the BDCU by the nurses in ED triage (however not registered in this table as diagnoses, but reasons for ED use).c The list of chronic physical illnesses is based on an adapted and validated version of the Elixhauser Comorbidity Index, integrating the Charlson Index, which consists of 32 major categories of physical illnesses (see reference in the Methods section).In this list of chronic physical illnesses, three categories of MDs and two of SRDs (identified with an asterisk [*]) were also included in the list of MDs-SRDs, thus appearing twice.
38ving in more materially and socially deprived areas: indexes 4-5 or areas not assigned b 55.38Violent/disturbed behaviors or social problems (measured from the first to the last ED use) a 30.25Clinicalcorrelates(measured from the first to the last ED use, or other as specified)High regularity of outpatient care (including usual GP, usual psychiatrist, and psychosocial clinicians from community healthcare centers) within the four 3-month periods of the year b,e 37.42Outcome (measured for the 12 months after the last ED visit)

Table 1 .
Cont.Maintenance et exploitation des données pour l'étude de la clientèle hospitalière (MED-ECHO-Hospital Inpatient and Day Surgery Database); e Système d'information permettant la gestion de l'information clinique et administrative dans le domaine de la santé et des services sociaux (I-CLSC-Community Healthcare Center Database); f Fichier des décès du registre des évènements démographiques (RED, Vital Statistics Death Database).

Table 2 .
Patient profiles based on their emergency department (ED) use (n = 11,682).

Table 3 .
Associations between patient profiles based on their emergency department (ED) use, sociodemographic and clinical characteristics, and service use, using multinomial regression (n = 11,682) (Reference: Profile 1).

Table 4 .
Associations between patient profiles based on their emergency department (ED) use and adverse outcomes, using Cox model (n = 11,682) (measured within 12 months after the last ED visit during the 2012-2016 period) (Reference: Profile 1).p-value < 0.05.HR = Hazard ratio.Each HR was adjusted for age and sex. *