Predictors of Frequent Recourse to Health Professionals by People With Severe Mental Disorders

Abstract

Objective:

Based on Andersen’s behavioural model, our study sought to determine predictors and blocks of factors that could explain why people with severe mental disorders (SMDs) more often seek the services of health professionals.

Methods:

This longitudinal study involved 292 users with SMDs located in Le Sud-Ouest, the southwest borough of Montreal. Data were collected from participants’ medical records and through 7 questionnaires. Using Andersen’s Behavioral Model of Health Services Use, independent variables were divided into 3 classes—predisposing factors, enabling factors, and need factors—and were introduced in this order in a hierarchical logistic model.

Results:

Among 292 users, 110 (37.7%) were frequent users who consulted about one health professional every 3 days. Participants who were more likely to call on health professionals were single and older, depended on welfare as their main source of income, lived in supervised housing, suffered from schizophrenia, schizophrenia spectrum disorders, and adjustment disorders, and, marginally, exhibited multiple mental disorders.

Conclusion:

Mental health services could promote strategies to overcome the reluctance of younger people to seek professional services. Professionals should pay close attention to subsidiary conditions, such as adjustment disorders, from which people with SMDs may suffer. Interventions to improve the socioeconomic condition of unemployed people with SMDs may help to reduce health care service use among that clientele. Programs such as supported employment should be tailored and enhanced for people receiving welfare to decrease stigmatization and improve job market integration.

Despite deinstitutionalization and the introduction of mental health services in the community during the last decades, few people with SMDs continue to be repeatedly admitted to specialized hospital services.¹ The disproportionate use of mental health services by frequent users (or heavy, high, or repeat users) has long been the focus of studies.^2,3 Some researchers have stressed that 5% to 18% of users with SMDs use between 27% and 63% of mental health services.^4–8 Other studies state that 30% to 35% of such patients account for 75% to 80% of psychiatric inpatient service costs.^9,10

Numerous studies during the last 2 decades have looked at factors of overuse of psychiatric emergency,^2,4,6,7,11 multiple readmission in specialized mental health care services (the revolving-door phenomenon),^3,12–15 and patterns of service use in general¹⁶ by people with SMDs. The leading variables identified as predictors of high service use by this clientele include sex,⁷ age,^7,17 marital status,⁴ unemployment,^4,6,15 education,^12,13 schizophrenia,^4,6,15 schizophrenia spectrum disorders,^13,15 substance use disorders,¹⁸ and lack of social support.^6,7,12 However, few studies have used a conceptual framework to try to understand the relative importance of those predictors regarding the high prevalence of health care service use.

The Behavioral Model of Health Services Use structured by Andersen¹⁹ is the most common theoretical framework used to identify factors of health care service use. According to Andersen’s theory, health care service use can be explained by 3 categories of variables (predisposing, enabling, and need factors). Predisposing factors are individual characteristics, such as age, sex, or level of education, that could prompt someone to use health care services. Enabling factors are systemic, organizational, and structural assets that could facilitate health care use, for example, insurance, family income, or employment. Finally, need factors are diagnoses or other clinical variables (such as severity of symptoms, chronicity, and perceived needs) that might justify a decision to seek help.⁸

Clinical Implications

• People who more rarely seek the services of health professionals are likely to be younger, married or in a relationship, to live in autonomous accommodations, and to have sources of income other than welfare.

• Among need factors that may lead participants to more frequently seek the support of health professionals are schizophrenia, schizophrenia spectrum disorders, adjustment disorders, and, marginally, multiplicity of mental disorders.

• Mental health care services should prioritize interventions that address modifiable factors, such as unemployment, that correlate with high service use.

Limitations

• There are differences between definitions of frequent and infrequent users according to the studies.

• Results may not apply in other geographical areas.

• Results may not be pertinent for people having no SMD.

In mental health, Andersen’s behavioural model has served primarily to identify determinants of health care services use for mental health reasons among people suffering mostly from common mental health disorders (anxiety or minor depressions).^8,20–27 However, the sole use of this model in relation to SMDs, to our knowledge, has been to identify patterns of medication use^28,29 or to predict health care service use by homeless people with SMDs.^30,31 All studies—with the exception of those by Lemming and Calsyn³⁰ and Rosenheck and Lam³¹—applied logistic regression to identify factors of mental health care service use and found that need factors were key predictors of service use. Hierarchical regression model analysis seems to be a more suitable method to account for patterns of mutually linked—compared with isolated—variables that explicate why some people with SMDs find it necessary to regularly seek the services of health professionals. The main purpose of our article is thus to determine predictors and blocks of factors that could explain frequent mental health service use, compared with infrequent use of services, by people with SMDs, based on Andersen’s theoretical model.

Methods

Study Design and Network Characteristics

This longitudinal study involved users with SMDs receiving treatment at an MHUI, located in the southwest borough of Montreal. This urban area, with a population of 258 000, comprises the MHUI and 2 HSSCs. The former offers specialized mental health services (that is, second- and third-line services), while the latter provide primary mental health care services. Other professionals and organizations offering health services in the area (for example, general practitioners, community-based services, and supervised housing) were discussed in more detail in previous publications.^27,32

Sample Selection Criteria and Recruitment of the Main Sample

Participants to the study were aged 18 to 65 and lived in the area served by the HSSCs. They exhibited an SMD based on diagnosis criterion 295 (schizophrenia and other psychotic disorders) or 296 (mood disorders) of the DSM-IV, and received T0 at the MHUI. They allowed the research team to see their medical record and to contact their principal case manager. The latter were subsequently asked to answer a questionnaire on their patient’s functional fitness. Candidates were excluded if they had been hospitalized, if they had severe mental retardation, or if they were committed to involuntary psychiatric treatment following a judicial decision—all conditions that would prevent these people from completing the questionnaires. Inclusion and exclusion criteria are comparable to those of previous studies that assessed the needs, quality of life, and patterns of health care service use among people with SMDs living in the community.^33–35

We conducted participant data collection between December 2008 and September 2010 (T0), and between January 2011 and November 2011 (T1). Each time, a trained interviewer met participants twice at a 1-week interval, with each interview lasting about 90 minutes. The ethics boards of the MHUI and the 2 HSSCs approved the study protocol.

Measurement Instruments

In addition to data collected from participants’ medical records at the MHUI, 7 questionnaires were administered in both English and French:

1. Service Utilization Questionnaire: adapted from the Statistic Canada’s Canadian Community Health Survey questionnaire,³⁶ it evaluates participant types and frequency of services used (for example, general practitioner and psychiatrist) for mental health reasons in a 12-month period preceding recruitment;

2. Alcohol Use Disorders Identification Test: measures the degree of addiction and at-risk alcohol consumption (Cronbach α = 0.88)³⁷;

3. Drug Abuse Screening Test: evaluates participants’ drug abuse and consequences (Cronbach α = 0.74)³⁸;

4. SPS: explores participants’ level of integration and social support (for example, reassurance about self-worth, need to feel useful; Cronbach α = 0.92)³⁹;

5. Alberta Continuity of Services Scale for Mental Health: measures service continuity (for example, system access and team function; Cronbach α = 0.78 to 0.92)⁴⁰;

6. MCAS: assesses patients’ functional status in the community (for example, obstacles to functioning, social competencies; Cronbach α = 0.87)⁴¹; and

7. MANQ: derived from the CAN.⁴² Four areas of need (stress adaptation, social inclusion, involvement in treatment decisions, and job integration) were added to the MANQ, in addition to the 22 areas assessed by the CAN, for a total of 26. The original CAN measures participants’ needs based on ordinal-scale questions, but the MANQ uses analogical scales, ranging from 0 to 10. The validation of the MANQ was the subject of a previous publication.⁴³

Participants answered all questionnaires except for the MCAS, which case managers completed. Participants’ medical records provided complementary clinical data: DSM-IV diagnoses (4 axes), history of suicide attempt and suicidal ideation, number of previous suicide attempts, history of violence, and history of legal problems.

Analyses

The analytical framework was built into the conceptual model displayed in Figure 1. The categorical dependent variable “High and low frequency of recourse to health professionals after 18-month follow-up period (T1)” was calculated by adding, for each participant, the number of visits with various health professionals (psychiatrists, family physicians, nurses, psychologists, health educators, social workers, and other professionals). Participants were then split into 2 groups using the mean value as the cutoff point. The rationale for dichotomizing the dependent variable was to contrast frequent with infrequent users, which could not be achieved with the continuous variable of frequency of recourse to a health care professional. Independent variables, as shown in Figure 1, were organized into 3 classes: predisposing factors, enabling factors, and need factors, according to Andersen’s Behavioral Model of Health Services Use.¹⁹

Figure 1.

Conceptual framework: predictors of frequent recourse to health professionals among people with severe mental health disorders

MANQ = Montreal Assessment and Needs Questionnaire; SUQ = Service Utilization Questionnaire

Variable blocks were entered in the model according to their antecedence in a participant’s social and medical history.³⁰ Predisposing factors usually came first, owing to their naturally occurring characteristics and strong links to the participant profile. Enabling factors were then entered to assess their effect on distal predisposing factors. Need factors are more proximal, and were thus entered as a final step as they are supposed to act as modifying factors on the conjoint effect of predisposing and enabling factors. The inclusion of variables into blocks was based on both the literature on health care service use and on our conceptual framework.^19,28–31

Analyses encompassed descriptive, bivariate, and hierarchical linear regression analyses. Descriptive analyses were carried out, first on the entire sample and then according to a comparison of frequent and infrequent users of services, along with comparative statistical tests to assess significant differences between the 2 groups. Differences were tested using the Pearson chi-squared test and the Fisher exact test for categorical independent variables, and the independent sample Student t test for continuous independent variables. Besides these comparison statistics, bivariate analyses allowed the selection of independent variables associated with the dependent variable, for an alpha value set at P < 0.10. Independent variables yielding a significant association were used to build a hierarchical logistic model, setting the alpha value at P < 0.05, and using a backward logistic regression elimination method. The total explained variance and the goodness of fit were generated.

Results

At T0, 437 participants were approached. A total of 352 subjects (81%) agreed to take part in the study, while 86 declined (20%). Among these 352 participants at T0, 297 (84%) agreed to respond to a T1. Refusals were compared with participants by age, sex, and housing at T0, and by age, sex, and diagnosis at T1. The comparison revealed no statistically significant difference between refusals and participants based on these variables at both T0 and T1.⁴⁴

On 297 participants at T1, 292 (98.3%) said that they had seen a health professional at least once during 12 months. Table 1 shows predisposing factors, enabling factors, and need factors. Regarding predisposing factors, the mean age for the 292 participants was 49 (SD 11), with 153 males (52%) and 139 females (48%). Most participants (64%) were French-speaking. Only 14% were married or living with a partner. Regarding enabling factors, 64% of participants were on welfare. The majority (63%) did not have a post-secondary diploma. Sixty-one per cent were living autonomously. Regarding need factors, the most prevalent SMDs were schizophrenia (38%) and mood disorders (43%).

Table 1.

Predisposing, enabling, and need factors associated with frequent recourse to health professionals, n = 292

Categories, variables		Level of health care service use
	Total sample n (%)	Frequent users n = 110 n (%)	Infrequent users n = 182 n (%)	P
Predisposing factors at T0
Age, mean (SD)	48.8 (10.5)	49.0 (9.5)	45.5 (10.9)	0.005a
Sex, n (%)				0.003b
Male	153 (52.4)	70 (63.6)	83 (45.6)
Female	139 (47.6)	40 (36.4)	99 (54.4)
Nationality, n (%)				0.74b
Canadian	266 (91.1)	101 (91.8)	165 (90.7)
Other	26 (8.9)	9 (8.2)	17 (9.3)
Spoken language, n (%)				0.009b
French	187 (64.0)	60 (54.5)	127 (69.8)
Other	105 (36.0)	50 (45.5)	55 (30.2)
Marital status, n (%)				<0.001c
Married or common law	40 (13.7)	3 (2.7)	37 (20.3)
Single	252 (86.3)	107 (97.3)	145 (79.7)
Importance attributed to spirituality, mean (SD)	7.4 (3.5)	7.2 (3.7)	7.6 (3.4)	0.24a
Level of help desired score, mean (SD)	49.8 (35.5)	53.0 (34.9)	48.0 (35.8)	0.23b
History of violence in past 12 months, n (%)	51 (17.5)	23 (20.9)	28 (15.4)	0.96a
History of legal problems in past 12 months, n (%)	28 (9.6)	7 (6.4)	21 (11.5)	0.15b
Enabling factors at T0
Source of income, n (%)				<0.001b
Welfare	186 (63.7)	91 (82.7)	95 (52.2)
Other sources	106 (36.3)	19 (17.3)	87 (47.8)
Education, n (%)				0.003b
Primary or high school	183 (62.7)	81 (73.6)	102 (56.0)
College or university	109 (37.3)	29 (26.4)	80 (44.0)
Type of housing, n (%)				<0.001b
Autonomous	177 (60.6)	42 (38.2)	135 (74.2)
Supervised	115 (39.4)	68 (61.8)	47 (25.8)
Number of kids, mean (SD)	0.6 (1.2)	0.5 (1.0)	0.7 (1.3)	0.06a
Number of family members giving help per month, mean (SD)	1.9 (1.8)	1.7 (1.7)	2.0 (1.8)	0.26a
Number of friends giving help per month, mean (SD)	1.6 (1.9)	1.5 (2.1)	1.6 (1.9)	0.62a
Frequency of contacts with caring family members per month, mean (SD)	31.6 (41.1)	22.3 (27.2)	37.2 (46.8)	0.003a
Frequency of contacts with caring friends per month, mean (SD)	26.4 (58.3)	24.8 (35.9)	27.4 (68.4)	0.71a
Help received from relatives score, mean (SD)	22.4 (23.4)	17.2 (22.1)	25.6 (23.6)	0.003a
Adequacy of help for self-perceived problems, mean (SD)	6.2 (2.4)	6.6 (2.5)	6.0 (2.4)	0.03a
Social Provisions Scale, mean (SD)	70.8 (7.3)	69.2 (7.5)	71.7 (7.0)	0.004a
Alberta Continuity of Services Scale for Mental Health, mean (SD)	134.3 (22.9)	134.4 (21.5)	134.2 (23.8)	0.96a
Need factors at T0
Mental disorders, n (%)
Severe mental disorders
Schizophrenia	112 (38.4)	66 (60.0)	46 (25.3)	<0.001b
Schizophrenia spectrum disorders	37 (12.7)	19 (17.3)	18 (9.9)	0.07b
Mood disorders, including mania and major depression	124 (42.5)	29 (26.4)	95 (52.2)	<0.001b
Delusion and other psychotic disorders	27 (9.2)	15 (13.6)	12 (6.6)	0.004b
Second diagnosis
Anxiety disorders	36 (12.3)	8 (7.3)	28 (15.4)	0.04b
Personality disorders	80 (27.4)	30 (27.3)	50 (27.5)	0.97b
Adjustment disorders	9 (3.1)	7 (6.4)	2 (1.1)	0.01c
History of suicide attempt(s), n (%)	93 (31.8)	31 (28.2)	62 (34.1)	0.30a
History of suicidal ideation, n (%)	112 (38.4)	37 (33.6)	75 (41.2)	0.30a
Number of suicide attempts, mean (SD)	0.7 (2.4)	0.8 (2.7)	0.6 (2.3)	0.66a
Number of mental health disorders, including substance abuse, mean (SD)	1.7 (0.9)	1.8 (0.9)	1.6 (0.8)	0.02a
Multnomah Community Ability Scale score, mean (SD)	65.4 (9.7)	63.2 (10.6)	66.7 (8.9)	0.003a
Alcohol Use Disorders Identification Test score, mean (SD)	6.0 (6.6)	6.2 (6.6)	5.9 (6.7)	0.78a
Drug Abuse Screening Test score, mean (SD)	2.3 (2.9)	2.3 (2.9)	2.3 (2.9)	0.93a
Number of self-perceived problems per participant, mean (SD)	7.9 (4.3)	7.6 (4.3)	8.1 (4.3)	0.28a
Perceived severity of problems by participants’ score, mean (SD)	50.2 (32.6)	49.1 (31.5)	50.8 (33.3)	0.65a
Health service use at T1
Total score of frequency of health service use at T1, mean (SD)	187.3 (211.0)
Heavy users of health services over total sample mean, n (%)	110 (37.7)	411.7 (183.3)	51.6 (47.3)

^aIndependent sample Student t test

^bPearson chi-squared test

^cFisher exact test

The fourth category of variables relates to health care service use. As it is the key dependent variable in our study, it appears in more detail in Table 1, which shows the distribution of participants’ characteristics according to frequency of service use (comparing frequent and infrequent users), along with statistically significant differences. The mean for number of visits to health professionals per year was 22.2 (equivalent to 1.8 visits per month). A total of 110 participants (37.7%) qualified as frequent users. On average, they consulted 1 health professional every 3 days (125.5 visits per year), while infrequent users visited a health professional once every 2 months or so (7.4 visits per year). Regarding predisposing factors, frequent users tended to be older, single males, and to speak a language other than French. Regarding enabling factors, this group was more likely to depend on welfare for income, to live in supervised housing, to be less educated, and, marginally, to have fewer kids. They had fewer contacts with caring family members on a monthly basis and received less help from relatives. They did receive more adequate support for their self-perceived problems, although their social support was not as strong according to the SPS score. Regarding need factors, frequent users were most likely to suffer from a broader range of mental health disorders, especially schizophrenia, mood disorders, delusions, and other psychotic disorders, and, marginally, schizophrenia spectrum disorders. They also tended to have received a secondary diagnosis of adjustment disorders and anxiety disorders. They were also less socially functional, based on the MCAS score.

Table 2 presents the hierarchical logistic regression model, comprising 3 blocks of variables related to predisposing, enabling, and need factors. Predisposing factors contributed to 20% of the total variance. It includes sex (male), French-speaking participants (negatively associated), marital status (single), and age (older). However, only age and marital status retained a significant beta in the final analysis. Enabling factors contributed to 13% of the total variance. Welfare and autonomous housing were retained and were found to have an impact on health professional use. Finally, need factors contributed to 8.5% of the total variance and included schizophrenia, schizophrenia spectrum disorders, adjustment disorders, and, marginally, number of mental health disorders. The goodness of fit for the 3 blocks was acceptable.

Table 2.

Predictors of frequent recourse to health professionals among participants with severe mental disorders: hierarchical logistic regression analysis, total sample, n = 292

	Model 1: Predisposing factors			Model 2: Enabling factors			Model 3: Need factors
Variable	P	OR	95% CI	P	OR	95% CI	β	SE	Wald	df	P	OR	95% CI
Sex (male)	0.01	1.952	1.152–3.306	0.07	1.688	0.962–2.963	0.423	0.310	1.853	1	0.17	1.526	0.830–2.804
French-speaking participants	0.005	0.470	0.276–0.800	0.29	0.726	0.400–1.317	−0.172	0.325	0.279	1	0.60	0.842	0.445–1.593
Marital status (single)	<0.001	9.189	2.682–31.482	0.008	5.527	1.551–19.694	1.289	0.654	3.883	1	0.049	3.628	1.007–13.076
Age	0.002	1.040	1.014–1.067	0.003	1.043	1.014–1.072	0.054	0.015	12.508	1	<0.001	1.056	1.024–1.088
Source of income (welfare)				<0.001	3.200	1.698–6.031	0.921	0.348	6.995	1	0.008	2.512	1.269–4.971
Housing (autonomous)				<0.001	0.303	0.169–0.540	−1.114	0.320	12.139	1	<0.001	0.328	0.175–0.614
Schizophrenia							0.927	0.345	7.228	1	0.007	2.527	1.286–4.968
Schizophrenia spectrum disorders							1.170	0.454	6.639	1	0.01	3.222	1.323–7.847
Adjustment disorders							3.023	1.026	8.682	1	0.003	20.553	2.752–153.529
Number of mental health disorders							0.300	0.175	2.943	1	0.09	1.350	0.958–1.903
Total explained variance
Nagelkerke, %	19.90	32.90	41.40
Goodness of fit
χ2, df = 1	5.632			6.827			2.026
P	0.69			0.56			0.98

Discussion

Our study shows that predisposing factors have the most impact on frequency of recourse to health professionals among people with SMDs, followed by enabling and need factors. Our results differ from those of the few previous studies that have used a hierarchical regression model and found that enabling factors contributed to a greater variance.^30,31 The number and nature of the selected variables included in the conceptual framework may explain these differences.⁴⁵ In our study, predisposing factors included several variables (spoken language, importance of spirituality, level of help desired, history of violence, and history of legal problem) that are either rarely or never analyzed. Another explanation would be that some variables usually included in a particular factor were classified differently by authors.⁴⁵ For example, Rosenheck and Lam³¹ placed social support among predisposing variables, while Lemming and Calsyn³⁰ included diagnoses in that category. Because of such inconsistencies, it is often difficult to compare the respective influence of predisposing, enabling, and need factors in the frequency of recourse to health professionals.⁴⁵

The order of introduction of predisposing, enabling, and need factors in the hierarchical regression model may also explain the differences. In Rosenheck and Lam’s study,³¹ predisposing and need factors were entered concurrently as a first step, followed by enabling factors. We also tested this latter model but it was our model that showed the greatest level of explained variance and had the most pertinent predictors. Finally, we need to consider that previous studies focused on homeless people with SMDs.^30,31 For these impoverished clientele, it may be that enabling factors were those that had the greater impact on access to health care services for mental health reasons.

Our results also confirm that people with SMDs who present less complex diagnoses are also less likely to use services. They have significantly fewer SMDs and secondary diagnoses; they are younger, married or in a relationship, live in autonomous housing, and have sources of income other than welfare. These variables usually portend a better quality of life and other favourable outcomes.^46–48

Concerning the link between frequent recourse to health professionals and specific predisposing factors among people with SMDs, our findings are in accordance with previously published data from similar studies.^3,4,7,17 Single or unattached people are more likely to use health care services for mental health reasons, a conclusion reported by several authors.^1,3,4,7 A spouse providing assistance often helps to reduce the need for mental health care services.⁴⁹ Conversely, mental health care services use can offset a lack of emotional or social support among people with SMDs.^7,50 Regarding age, older people with SMDs are more likely than the general population to have occurring comorbid physical conditions and, consequently, to use health care services for mental health reasons more often.⁵¹ According to some authors,^52–55 50% to 90% of people with SMDs may also suffer from physical problems (for example, hypertension, diabetes, cancer, and dental problems). By some estimations, the risk of dying from respiratory or heart diseases may be 4 times higher among that group, compared with the general population.^55,56 Meanwhile, younger people tend to be more reluctant to seek health care services for mental health reasons in the first place, owing to mistrust of medical personnel, doubts about the effectiveness of treatment, financial barriers, or because they prefer to solve their problems by themselves.⁵⁷ Unsurprisingly, younger people are also more likely to abandon treatment,⁵⁷ if they ever reach that step.

Concerning enabling factors, it makes sense that people living in autonomous housing would require fewer services from health professionals.⁵⁸ According to the residential continuum model dominant in mental health, people with SMDs move to more autonomous housing when their condition stabilizes and requires less professional support. Meanwhile, people living in supervised housing have more opportunity to be visited by their case manager or other health professional. People with SMDs who receive welfare are also more likely to overuse health care services for mental health reasons, a fact reported by many authors.^4,15,59 Poverty causes stress, psychological distress, and depression, and that could explain the frequent recourse to health professionals.^16,60 Moreover, people with SMDs are often victims of stigmatization and face serious obstacles in accessing the labour market or in finding an adequate apartment.^61,62 The likelihood of relapse and readmission to hospital also makes it difficult for them to keep a job.³ Finally, people with SMDs often have a poor quality of life and low self-esteem and tend to self-stigmatize.^63,64 Having a job and access to autonomous housing are important factors to reduce the dependence of people with SMDs on health and social services and to promote their social integration and recovery.^65,66

While need factors rank third in terms of explained variance, some variables produce the strongest association, with frequent recourse to health professionals, as the plots for adjustment disorders and schizophrenia spectrum disorders show. SMDs, such as schizophrenia and schizophrenia spectrum disorders, understandably comprise a factor of frequent reliance on health professionals.^1,3,5,13,15 One reason might be that people suffering from SMDs need more follow-up by an assertive community team or case manager, owing to their higher rate of relapse.^67,68 Carr et al¹⁶ found that people with schizophrenia and other SMDs who had been hospitalized in the year prior to their interview had, on average, one contact with a health professional per week. Another explanation would be people with SMDs have serious problems in several areas (for example, psychiatric symptoms, physical health, daytime activities, company, and food), and, therefore, require the services of various health professionals.⁶⁹ Meanwhile, adjustment disorders often lead to dramatic and manifest crises that can be highly stressful for relatives, owing to the risk of harm to the patient or others.⁷⁰ We reported earlier on the marginal link between the number of mental health disorders and the high incidence of health professional use by people with SMDs.⁴ The higher risk of relapse among people suffering from comorbid mental health disorders may explain this link.

Limitations

Some limitations to our study should be elaborated on. First, the definition of frequent and infrequent users varies from one study to another, which could explain some discrepancies between our main findings and those of other studies. Second, the Canadian context of this study, and its confined scope within a particular borough in Montreal, may limit the use of its findings. Third, some participants who qualified as infrequent users could have been frequent users during short periods. Fourth, professional services offered by some community-based organizations (mainly follow-up and treatment in the community) were excluded from the analysis. Fifth, data on health care professional use were self-reported. However, we used the same instrument as many previous studies^71,72 that have stressed the accuracy of self-reported data on this topic.^73–75 Finally, our results may not be applicable to populations having common mental health disorders or to people with a specific SMD, such as schizophrenia only.

Conclusion

Our study is one of the few that have used Andersen’s behavioural model in combination with hierarchical regression to determine blocks of predictors of frequent recourse to health professionals among people with SMDs. Its prospective design brings out respective importance of predisposing, enabling, and need factors that cause people with SMDs to frequently rely on the services of health professionals. The knowledge of predictors of each of those factors could be useful for clinicians or decision makers. Predisposing factors help to identify groups (for example, younger people) who are more reluctant to use mental health services. Strategies (for example, outreach) should be promoted to help change the attitude of people who infrequently use services even when they need them. Need factors indicate that schizophrenia and schizophrenia spectrum disorders are the SMDs most often linked with high health care services use for mental health reasons. Frequent users with SMDs also tend to suffer from other mental disorders. Professionals should pay close attention to subsidiary conditions, such as adjustment disorders, from which people with SMDs may suffer. Enabling factors are the more easily modifiable. Mental health care services should not only serve to alleviate suffering from psychotic crises but also address socioeconomic challenges and modifiable factors, such as unemployment, that correlate with overuse of services. People with SMDs who receive welfare are frequent users of health care services, which justify a greater effort to help this clientele. Programs, such as supported employment, should be tailored and enhanced for this specific group to decrease stigmatization and improve job market integration. However, further studies are needed to broaden the range of cost-efficient determinants that could be successfully targeted by such interventions.

Abbreviations

CAN

Camberwell Assessment of Needs

DSM-IV

Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition

HSSC

health and social service centre

MANQ

Montreal Assessment of Needs Questionnaire

MCAS

Multnomah Community Ability Scale

MHUI

Mental Health University Institute

SMD

severe mental disorder

SPS

Social Provisions Scale

T0

treatment at baseline

T1

treatment at 18-month follow-up