Background

Quality of life (QoL) is a key outcome in the planning and evaluation of health services, notably for patients with chronic illnesses or vulnerable populations – e.g., patients with mental disorders (MD), who often are frequent emergency department (ED) users, with 4 + ED visits/year [1,2,3]. According to the World Health Organization (WHO), QoL is a multidimensional concept that integrates various aspects of health and well-being in relation to the individual’s environment and beliefs [4]. QoL is strongly associated with the recovery movement [5] of patients achieving a satisfying life in spite of their illnesses or vulnerabilities [6]. Well-being, engagement in meaningful activities, low self-stigma, and the feeling of belonging to a community are characteristics of high QoL among patients with MD, including substance-related disorders (SRD) [7, 8]. QoL may also be used as a proxy for measuring the met needs of patients, access to care, or satisfaction with services [9, 10]. Patients using ED, and frequent ED users in particular, are often reported as having high unmet needs [11] and barriers to care [12], a consequence of usually insufficient access to and adequacy of outpatient care [13]. This suggests that high ED users could have low QoL. A better knowledge of QoL, sociodemographic and clinical profiles of ED users with MD, associated with outpatient service use and satisfaction, may sustain recommendations to improve health services for these vulnerable patients.

Though numerous studies investigate determinants of QoL among patients with MD [14,15,16,17,18,19,20], few have examined QoL profiles [21, 22]. In fact, we found no previous study investigating QoL profiles among ED users. While few published studies have identified ED user profiles among patients with MD, several have assessed determinants of ED use [23, 24] or high ED use [25,26,27,28,29,30,31] among such patients. A previous study [32] identified four QoL profiles among patients with MD; the profile with the best QoL was of older men with low education, serious MD, few important needs, and receiving a high level of help from health services; whereas the profile with the worst QoL included young women with co-occurring MD-SRD and important needs [32]. Yet another [33] reported four profiles of ED users with MD, with the one with the highest ED use also having the highest specialized service use among patients with multiple MD-SRD. Another recent study [34] showed three profiles of high ED users with MD: (1) 3-year recurrent very high ED users (10 + ED visits/year); (2) 2-year recurrent high ED users; (3) 1-year high ED users. Profiles differed in their severity of health conditions, intensity of outpatient service use, and risk of death, all of which were high in Profile 1, moderate in Profile 2, and low in Profile 3.

QoL studies usually focus on patients with specific MD: serious MD [14], bipolar disorders [15, 16], depressive disorders [17,18,19], borderline personality disorders [20], and SRD [35]. Studies found that QoL was mostly lower among women [15], patients with several MD [15, 20], general anxiety disorders [17], drug-related disorders [14], or more severe depressive symptoms [15, 16, 18, 20]. Conversely, individuals in a relationship [19], and who had better social integration [14], self-efficacity [16], schizophrenia [14], and few important needs [14] reported higher QoL. Based mainly on patient medical records, determinants of high ED use were mostly associated with being male [36, 37], younger [28, 29], poor [30], and diagnosed with schizophrenia [25, 31], SRD [31], or personality disorders [26, 27].

Investigating QoL profiles among ED users in association with their service use and level of satisfaction may provide complementary information that could enable health practitioners to better respond to the comprehensive needs of these patients, and decrease their high ED use. This study is original in that it investigates QoL among ED users based on both patient surveys and medical records, including a broad set of variables related to patient characteristics and service use data. Based on 299 patients who used different ED health networks for mental health (MH) reasons in Quebec (Canada), this study aimed to identify patient profiles based on their QoL, sociodemographic and clinical characteristics, and associated these profiles with patient service use. Identifying QoL profiles in this way might help decision makers deploy interventions that are more adapted to each profile, and provide more individualized treatment.

Methods

Study setting and data collection

The study was conducted in four Quebec (Canada) ED networks serving a population of about two million individuals – roughly one fourth of the province’s population. The first ED was in a large Montreal psychiatric hospital, while the second was also an ED in a psychiatric hospital but merged within a general hospital. Most patients had to go to the ED of the general hospital, from where they were referred to the psychiatric ED. The third was a psychiatric ED located in a large university-affiliated general hospital in Quebec City, where patients first went before being transferred to the psychiatric ED. The fourth ED, located in an outlying area of Montreal, was integrated within a general hospital where the staff included psychiatric clinicians.

Study participants had to be ED users for MH reasons, diagnosed with MD including SRD, 18 + years old, able to complete a survey, French or English speakers, and had to grant the research team access to their medical records within the targeted networks. Recruitment was conducted randomly from a list of 1,751 ED users with MD, identified as such in the medical records filled by ED clinicians from these networks. The coordinates of the first 450 patients that ED clinicians identified as interested in participating in the study were referred to the research team as potential study participants. The research coordinator then contacted these individuals, asking them to take part in a survey conducted by trained interviewers, with close monitoring from the research team. The surveys were conducted by telephone, due to restrictions regarding face-to-face contact during the COVID-19 pandemic. They were administered between March 1st, 2021 and May 13th, 2022, each interview lasting around 45 min. As the surveys went on, medical records data were collected for the 12 months preceding patient interviews, except for recurrent frequent ED use, which was measured for the 2 years preceding the 12 months of ED use targeted in the interviews. Medical records included information about ED use (BDCU database; e.g., date, reason for visit – Banque de données communes des urgences) [38] and hospitalization for MH reasons (MED-ECHO database; e.g., dates of admission/discharge, duration of inpatient stay – Maintenance et exploitation des données pour l’étude de la clientele hospitalière) [39], psychiatric outpatient services used (hospital database; e.g., date, type of services received), and psychosocial MH services received from community healthcare centers (I-CLSC database; e.g., date, type of services received – Système d’information sur la clientèle et les services des centres locaux de services communautaires) [40]. Patient diagnoses were based on the BDCU and MED-ECHO databases, framed by the International Classification of Diseases, Tenth Revision (Appendix 1). All databases included information on patient service use (e.g., frequency) exclusively within the ED networks. A steering committee including ED clinicians validated the survey data to be collected, including data on integrated service use outside the ED networks and data on services other than those in network hospitals and community healthcare centers (e.g., medical clinics, community-based organizations), which completed the information taken from medical records. The variables extracted from the survey data and individual medical records are presented in a conceptual framework (Fig. 1) including the standardized scales used. Participation in the study was voluntary. Patients provided oral consent and received a modest financial compensation. The ethics review board of the Douglas Mental Health University Institute approved the multi-site protocol.

Fig. 1
figure 1

Analytical framework: Profiles of quality of life (QoL) among patients using emergency departments (ED) for mental health (MH) reasons (N = 299). a Satisfaction with Life Domains Scale (included in the survey); b Survey/questionnaire; cBanque de données communes des urgences (BDCU, ED database); dMaintenance et exploitation des données pour l’étude de la clientèle hospitalière (MED-ECHO, hospitalization database); eAlcohol Use Disorders Identification Test (AUDIT, included in the survey); fDrug Abuse Screening Test-20 (DAST-20, included in the survey); gSystème d’information permettant la gestion de l’information clinique et administrative dans le domaine de la santé et des services sociaux (I-CLSC, community health center database); h “Psychiatric outpatient services used” database. For the definitions of variables included in the study, see the Table 1 footnotes or the Methods section. Details for the diagnostic codes are presented in Appendix 1

Full size image

Study variables

The dependent variable, QoL, was assessed using the Satisfaction with Life Domains Scale, a 20-item standardized questionnaire based on a 7-point scale, with higher scores indicating better QoL [41, 42]. Apart from QoL, patient profiles included sociodemographic and clinical variables, which were then associated with service use to differentiate patients with MD using ED.

Sociodemographic characteristics included: sex, age group, education (e.g., high school or less), employment status (e.g., worker, unemployed), household income (CAN$), type of housing (e.g., supervised), and level of stigma. In line with the Canadian Community Health Survey [43] and previous studies that have assessed stigma using a one-item question [44, 45], stigma was measured using the following question, with responses on a 5-point scale ranging from “totally disagree” to “totally agree”: Most people in my community treat a person with a MD or SRD in the same manner as they would any other person. Responses of “totally disagree” and “somewhat disagree” represented high stigma.

Clinical characteristics included: MD, suicidal behaviors (suicide ideation or attempt), good perceived mental/physical conditions, and chronic physical illnesses (e.g.: heart diseases, diabetes). MD included serious MD (schizophrenia spectrum and other psychotic disorders, bipolar disorders), personality disorders, common MD (anxiety, depressive and adjustment disorders, attention deficit/hyperactivity disorder), and SRD integrating alcohol- and drug-related disorders (use, induced, intoxication and withdrawal). SRD were measured using medical records. Considering that SRD are often under-diagnosed [46], the Alcohol Use Disorders Identification Test [47] and the Drug Abuse Screening Test-20 [48] were also used. The list of chronic physical illnesses was based on an adapted version integrating both the Charlson and Elixhauser Comorbidity indexes [49]. Perceived mental/physical health conditions were measured with two questions based on the Canadian Community Health Survey [43]. The questions (How do you see your MH? How do you see your physical health?) were on a 10-point scale ranging from poor to excellent, with 7 + representing good perceived mental/physical health conditions.

Service use included: knowledge of MH services; frequent outpatient MH service use; frequent ED use, recurrent frequent ED use, and hospitalization for MH reasons (yes/no); adequacy of care; and satisfaction with overall service use. Knowledge of MH services was measured with the following question: How do you evaluate your knowledge of MH or addiction services? This question on a 4-point scale ranged from poor or fair to excellent, with 3 + considered very good to excellent knowledge. Outpatient MH services included consultations with general practitioners (family doctors or GP from walk-in clinics), in community healthcare centers providing mostly psychosocial services, and with psychologists in private practice. They also included services from psychiatrists, most of whom (about 90% of Quebec psychiatrists  [50]) worked in hospital multidisciplinary teams, assertive community treatment and intensive case management programs, addiction rehabilitation centers, and community-based organizations (e.g., crisis or suicide prevention centers). A benchmark of 30 + services/year was established for frequent outpatient service use, representing roughly 2.5 + services/month, or 3 + over a 10-month period. An ED use frequency of 4 + visits/year is a standard definition used in Canadian studies [51, 52]. Recurrent frequent ED use was defined as 8 + visits over the 2 years preceding the 12 months considered in the surveys. Previous studies have shown that frequent ED users often use ED for several consecutive years, thus contributing to the “revolving door syndrome” [53, 54]. Adequacy of care was measured with a question with a 5-point response scale, from the Canadian Community Health Survey [43]: Do services outside of ED respond to your needs? Answers included “(1) I totally disagree”, “(4) I agree”, and “(5) I totally agree”, with good adequacy of care corresponding to answers 4 and 5. The service satisfaction variable considered for each service used by patients was measured on a 5-point scale ranging from “(1) Not at all satisfied” to “(5) Totally satisfied”. The mean level of satisfaction with services used per patient was also considered, with higher scores indicating greater satisfaction.

Analysis

Missing values (less than 1%) were randomly distributed and imputed by mean and mode [55]. Descriptive analyses were produced by frequency distributions for categorical variables, and mean values with standard deviations (SD.) for continuous variables. Cluster analysis was conducted using the k-means group algorithm with Gower dissimilarity coefficient based on QoL, sociodemographic and clinical characteristics. Several k-means solutions with different numbers (3–7) of profiles were computed [56]. The four-group solution had the largest Calinski–Harabasz pseudo-F value [57], indicating that it was the most distinct compared to the other models. Comparison analyses were conducted to assess statistical differences between profiles, using Chi-square or Fisher’s exact tests for categorical variables, and T-tests for continuous variables. Analyses were performed using Stata 17 [58].

Results

Of the 450 ED users referred to the research team, 50 could not be reached and 300 accepted to participate in the study (75% response rate). One patient was withdrawn as data collection was completed twice for this patient in two ED networks. The mean QoL score for the 299 patients in the final sample was 4.55/7 (Table 1). Most patients were 30 + years old (69%), 55% were women, 57% unemployed or retired, 57% had some post-secondary education, 22% lived in supervised housing, 50% had high perceived stigma, and 47% had a household income of less than CAN$20,000/year. Most (59%) reported having SRD, 57% common MD, 45% serious MD, 42% personality disorders, 54% suicidal behaviors, 45% chronic physical illnesses, and 32% perceived their mental/physical health conditions as good (7+/10). As for service use, 54% reported having good knowledge of MH services, 51% showed frequent outpatient service use (30 + times/year), 61% frequent ED use, 39% recurrent frequent ED use, 56% were hospitalized, and 63% had received adequate care. The mean score for overall service use satisfaction was 3.96/5.0.

Table 1 Characteristics of patients (N = 299)
Full size table

Cluster analysis identified four groups with quite different QoL scores and sample sizes. The superscript numbers in Table 2 indicate significant differences between profiles, as reported below. Accounting for 25% of the sample, Profile 1 had the lowest QoL score (3.99/7). All Profile 1 patients were men, differing as such from Profiles 3 and 4 which included few men (0% and 33%, respectively). Profile 1 had the most patients who were unemployed or retired (82%) and showed the lowest household income of all profiles, with 73% reporting earnings of less than CAN$20,000/year. Profile 1 also had more patients with high school education or less compared to Profile 2 (27% vs. 6%), and more living in supervised housing (55% vs. 31%). Profile 1 patients were more affected by SRD than those of Profile 4 (69% vs. 51%), but less affected than Profile 3 by personality disorders (31% vs. 57%) and common MD (47% vs. 67%). Only 11% of Profile 1 patients perceived having good mental/physical health conditions, many fewer than in Profiles 4 and 2 (100% and 31%, respectively), but still more than Profile 3 (0%). Profile 1 was labeled: “Unemployed or retired men with low QoL, education and household income, mostly having SRD and bad perceived mental/physical health conditions”.

Table 2 Characteristics and comparisons between profiles of the four-group model (N = 299)
Full size table

Accounting for 12% of the sample, Profile 2 had the highest QoL score along with Profile 4 (4.99 vs. 5.25/7, so not significantly different). Like Profile 1, Profile 2 only included men. Out of the four profiles, Profile 2 patients showed the highest household income, with earnings of CAN$40,000+/year. They were workers or students in higher numbers (71%) than in Profiles 1 and 3 (18%, 43%), and more of them lived in private or rented housing (94%) compared to Profiles 1 and 3 (73%, 75%). More Profile 2 patients had a post-secondary education compared to Profile 1 (69% vs. 46%). They were less affected by personality disorders (20%) than those of Profiles 3 and 4 (57%, 42%), and showed less suicidal behaviors than Profile 3 (40% vs. 61%). Profile 2 ranked second in perceived mental/physical health conditions, with 31% perceiving good conditions. Profile 2 was labeled: “Men who are employed or students, have good QoL, high education and household income, the least personality disorders, and fair perceived mental/physical health conditions”.

Accounting for 38% of the sample, Profile 3 had the lowest QoL score (3.99/7) after Profile 1. All Profile 3 patients were women, and none perceived having good mental/physical health conditions. Most (51%) reported a household income below CAN$20,000/year, with 57% of them being unemployed or retired – both figures lower here than in Profile 1 (73%, 82%), but higher than Profile 2 (0%, 29%). Profile 3 patients were more likely to live in supervised housing than those in Profile 2 (25% vs. 6%). Most were affected by personality disorders (57%), more had common MD than in Profile 1 (67% vs. 47%) and suicidal behaviors than in Profile 2 (61% vs. 40%), but fewer had serious MD compared to Profile 4 (39% vs. 55%). Profile 3 was labeled: “Women with low QoL, multiple MH problems, and very bad perceived mental/physical health conditions”.

Accounting for 25% of the sample, Profile 4 had the highest QoL score (5.25/7), although comparable to Profile 2 (4.99). Most were women (67%), all perceiving their mental/physical health conditions as good. Of all Profile 4 patients, 38% reported a household income of less than CAN$20,000/year; by contrast 51% of Profile 3, 73% of Profile 1, and 0% of Profile 2 were at that income level. With 43% unemployed or retired, Profile 4 patients differed from those of Profile 1, where 82% were working or studying. Profile 4 included more patients with personality disorders (42%) than in Profile 2 (20%), but fewer than in Profile 3 (57%). Profile 4 also had more patients with serious MD than Profile 3 (55% vs. 39%), but fewer with SRD than in Profile 1 (51% vs. 69%). Profile 4 was labeled: “Mostly women with very good QoL, serious MD, and very good perceived mental/physical health conditions”.

Patient profiles also differed regarding service use (Table 3). Profile 4 patients showed the highest satisfaction with their overall service use (4.27/5), had better knowledge of MH services (66%) than Profiles 1 and 2 (both 46%), and reported higher adequacy of care than Profile 3 (71% vs. 56%). Profile 3 patients, on the other hand, accounted for more frequent ED use and outpatient service use (70%, 60%) than Profiles 1 (69%, 46%) and 4 (72%, 55%). Recurrent frequent ED use was higher in Profile 3 (49%) than in Profiles 2 and 4 (26%, 30%).

Table 3 Associations between profiles and service use (N = 299)
Full size table

Discussion

This study identified four QoL patient profiles among ED users based on their sociodemographic and clinical characteristics, and associated these profiles with their service use. The study patients’ mean QoL score was relatively low (4.55/7), which could be explained by their overall high prevalence of social and health problems. Roughly half were unemployed or retired, had low household income, high perceived stigma, serious MD, SRD, personality disorders, suicidal behaviors or co-occurring chronic physical illnesses. Over 20% of them were living in supervised housing. Overall, study patients were also frequent service users, with most showing frequent ED use: over 30% reported frequent ED use for 3 consecutive years, and more than 50% had been hospitalized for MH reasons.

Profiles 2 and 4 had the best QoL, with two-thirds of the latter composed of women and all of the former consisting of men; although it should be noted that Profile 2 accounted for only 12% of the sample. All Profile 4 patients perceived their mental/physical health conditions as good, as did a third of those in Profile 2, which ranked second in this respect. Profile 4 included the most patients with serious MD, followed by Profile 2, which had the least patients with personality disorders. Self-perceived well-being is a key predictor of QoL [8] and a positive health outcome [59]. Studies have shown that patients with serious MD have better QoL than those with other MD [60] or low income [61]. The recovery movement may have helped patients with serious MD to lead a good life despite their chronic illnesses. These patients may also have entertained fewer expectations, especially if they received the help they needed. Like patients with personality disorders, they are known to have low QoL [62]. Profile 2 had the best social conditions, with Profile 4 a close second. These two profiles had the most patients working or studying, those with more household income and fewer (especially Profile 2) living in supervised housing. Indeed, previous studies have found that having a job increases self-esteem [63], reduces the need for treatment [63], and has a positive impact on health-related QoL [64], especially among patients with serious MD. Study patients usually preferred to live autonomously whenever possible [65], which may explain higher QoL in Profile 2 and 4 patients, who lived mainly in private or rented housing. Profile 4 patients were the most satisfied with services received, and, along with Profile 2, reported the highest adequacy of care. Profile 4 patients also had the best knowledge of MH services, which may have helped them get the help they needed. Higher appreciation of services in Profiles 4 and 2 may have contributed to their better overall situation and QoL compared to Profiles 1 and 3.

Profile 1 patients, who were all men, had the lowest QoL, the most disadvantaged social characteristics, and were the most affected by SRD, with fewer perceiving their mental/physical conditions as good. While half of patients in Profile 3 reported recurrent frequent ED use, those in Profile 1 also scored high in this regard, with most being unemployed and having low household income (<$20,000). The negative associations between poverty, unemployment, poor perceived health conditions and QoL have often been reported [66, 67], and thus, contrary to the men in Profile 2, about a third of Profile 1 patients were living in supervised housing, which may have contributed to their lower QoL scores. Previous studies have shown that QoL may be lower among patients living in such settings, compared to those living autonomously [65]. Previous studies also found poor QoL among patients with SRD [35, 68], which was associated with loss of a job or relationship [69, 70], higher morbidity [71], and affected many more men than women [72]. Men and patients with SRD are known to be frequent ED users [73, 74]. Compared with women, men tend to use health services as a last resort [75]. Patients like those in Profile 1, who make high use of ED over several consecutive years, are also known to be a vulnerable group. This pattern, compounded by the cumulative impact of deprivation and low health perception, may explain why Profile 1 showed the worst QoL.

Profile 3 was second worst in QoL, with patients sharing the same disadvantaged social characteristics found in Profile 1 (e.g., unemployment, low income, living in supervised housing), with the difference that all Profile 3 patients were women. They were the most affected by personality disorders, common MD, and suicidal behaviors, and were the most frequent service users, none of whom perceived their mental/physical conditions as good. In our study as in the existing literature, patients with personality disorders, and particularly borderline personality disorders, are found to have great difficulty maintaining jobs and relationships [76, 77], more negative perceptions of their physical mental/conditions [78, 79], and more likely to be high service users [80]. The greater prevalence of common MD compared to Profile 1, and of suicidal behaviors compared to Profile 2, may also be related to personality disorders, as those are often associated [81]; this prevalence may explain why Profile 3 accounted for nearly 40% of the sample. Suicidal behaviors, for their part, have quite logically been negatively associated with QoL [82]. Considering their numerous social and health issues, it is no surprise that Profile 3 patients are known to be high ED [83] and outpatient care users [80]. These characteristics may explain why they had the second worst QoL after Profile 1.

Limitations

This study has some limitations. First, patients were recruited from ED with psychiatric care in large urban areas, operating within a public healthcare system, so findings may not be generalizable to other types of ED, territories, or contexts. Second, only overall QoL was measured. If some aspects of QoL (e.g., health, living situation) had been measured separately, they might have been more strongly associated with specific patient characteristics and patterns of service use. Third, as the study was cross-sectional, the causality between associations could not be established. Fourth, outpatient MH care was assigned only to one variable. Finer details about services could have enabled us to further distinguish quality of care between profiles. Finally, it was impossible to test how the COVID-19 pandemic may have affected the QoL of ED users. Data collection for the study did not however occur during the first waves of the COVID-19 pandemic, which had the worse impact on ED users [84].

Conclusion

This study was original in identifying four distinct QoL profiles among patients who were high ED users: Profiles 4 and 2 reported high QoL; Profile 1, low QoL; and Profile 3, moderate QoL. Men and women were always contrasting in high QoL as opposed to low/moderate QoL profiles, with Profiles 1 and 2 including only men, Profile 3 only women, and two-thirds of Profile 4 comprised of women. The study showed that profiles with the highest QoL (4 and 2) also had better social characteristics and perceived health conditions. And while Profile 4 had the best QoL (although not significantly higher than Profile 2), it also registered the highest level of satisfaction with services received. Consequently, interventions aimed at improving QoL may be recommended to enhance the social conditions of patients, especially employment and household income. Programs like permanent supportive housing, and individual placement and support might be suggested, especially for Profiles 1 and 3. In their practices, clinicians could be more attentive to patients, and integrate more individualized treatments to better respond to their patients’ perceived needs. The recovery approach could be promoted more, and alliances made with patients for implementing more adequate care. Satisfaction with care, found to be a promising measure associated with high QoL, might be more routinely assessed within the healthcare system to benefit patients.