Predictors of Number of Healthcare Professionals Consulted by Individuals with Mental Disorders or High Psychological Distress

This study assesses the contribution of predisposing, enabling, and needs factors and related variables that predicted the number of healthcare professionals consulted for mental health reasons among 746 individuals with mental disorders and high psychological distress. The data were drawn from the third (T3) and fourth data collection periods (T4) of a longitudinal study conducted in a Quebec/Canada epidemiological catchment area. Hierarchical linear regression was performed on the number of types of healthcare professionals consulted in the 12 months prior to T4. Predictors were identified at T3, classified as predisposing, enabling, and needs factors (i.e., clinical and related variables) according to the Andersen Behavioral Model. Three needs factors were associated with the number of types of healthcare professionals consulted: Post-traumatic stress disorder, stressful events, and marginally suicide ideation. Three enabling factors: Having a family physician, previous use of mental health services, and employment status were also related to the dependent variable. Poor self-perception of mental health status was the only predisposing factor retained. While needs factors were the main predictors of the number of types of healthcare professionals consulted, enabling factors may reduce the influence of needs factors, by the deployment of various strategies that facilitate continuous and appropriate care.


Introduction
Mental disorders (MD) represent a challenge to the healthcare system due to their high prevalence, costs, and burden of disability, while negatively impacting quality of life for affected individuals and families [1,2]. Recurring MD and physical comorbidity are also common [3,4], leading many individuals to seek extensive and diversified care [5]. According to previous studies, as many as 70% of individuals affected by MD who use services consult professionals in primary care settings for prevalent mental health (MH) conditions such as anxiety, mood, and substance use disorders [6,7], while approximately 36% of individuals with MD seek care from general practitioners and MH specialists [5]. General practitioners at the entry point to primary care are largely responsible for care co-ordination and patterns of service use. Patients perceive help received from general practitioners in combination with MH specialists as highly effective [8]. For instance, psychotherapy provided by a psychologist is commonly recommended for the treatment of common MD (e.g., depression, anxiety), with medication provided by a general practitioner [9]. Case managers, often nurses or social disorders, which caused suffering, psychosocial, and work-related impairment. They also were at a greater risk for developing threshold GAD and other anxiety, mood, and substance use disorders [29]. In addition, having an undiagnosed MD was a significant predictor of lifetime hospitalization, lifetime MD without current symptoms, suicide attempts, high psychological distress, and other significant disability [27].
In order to form a better picture of individuals in need of MH services, it may be useful to investigate MH service utilization among individuals without a diagnosis, but who are at high risk for MD [30]. Accordingly, this study focused on both individuals with a diagnosed MD and those with high psychological distress as measured by: (1) Symptoms similar to those associated with common MD including depression and anxiety [31]; (2) conditions likely to evolve into MD if left untreated; and (3) conditions associated with high levels of functional impairment [32]. MH service utilization studies that include participants with high psychological distress at risk for developing MD, as well as diagnosed individuals, also conform to MH policies encouraging early intervention [33,34]. Thus, using the Andersen model, this study aims: (1) To assess the relative contribution of predisposing, enabling, and needs factors on the number of types of healthcare professionals consulted for MH reasons and (2) to identify predictors of a greater diversity of healthcare professionals consulted in a sample of both individuals with diagnosed MD and those with high psychological distress. As needs factors were mainly associated with health service use in previous studies, we hypothesized that individuals with more serious needs would seek help from a greater variety of MH professionals.

Study Design and Setting
This study is based on a subsample from a longitudinal population-based cohort study, conducted in an epidemiological catchment area in Southwest Montreal, Quebec (Canada), with a population of 269,720 distributed among six neighborhoods. Data for this specific study (the subsample) included the two last data collections from the full longitudinal study (T3, T4). The area includes a major psychiatric hospital offering specialized services for serious and complex MD, and two community health and social service centers that deliver primary MH care. Numerous private clinics employ general practitioners and psychologists, while community organizations offer supportive services such as crisis centers, day centers, and self-help groups for individuals with MH problems and their families.

Selection Criteria and Study Population
Study participants had to be between 15 and 65 years of age and live within the catchment area. A geographically representative sample with socio-economic and other characteristics proportional to those of the general population was sought. A random sample of 3408 home addresses was initially selected for recruitment purposes. In order to improve the recruitment process, this initial list was extended to include a range of 14 neighboring addresses for door-to-door recruitment [35]. Interviews were scheduled with individuals who agreed to participate either at home or at another designated location. Trained interviewers administered questionnaires to participants. The duration of interviews ranged from 90 to 135 min depending on whether or not participants had positive signs of MD and/or used MH services. Each participant signed a consent form prior to the interview. The anonymity of participants was guaranteed, and data were kept in a place accessible only to project researchers. Telephone contacts were made every six months encouraging participants to participate in the subsequent data collection periods. The ethics review board of a MH institute approved the multi-site protocol.
The first data collection period (T1: June 2007-December 2008) included N = 2434 study participants. Three additional data collection periods followed at two-year intervals (T2: June 2009-December 2010; T3: June 2011-December 2012; T4: June 2013-December 2015); they are summarized in Figure 1. The response rate for this longitudinal cohort was: 74.9% at T2, 72% at T3, and 80% at T4. These response rates were slightly higher than those in comparable longitudinal epidemiological studies (69% to~76% for other two to five year studies) [36,37]. Most dropout between data collection periods involved lower participation by younger or materially deprived individuals and those diagnosed with substance use disorders. Further details on selection criteria and sample characteristics have been published elsewhere [5,35]. At the fourth data collection period, the mean age of participants (N = 1871) was 44.72 years (SD: 13.86); 60% were female, and a small majority lived alone (52%). collection periods involved lower participation by younger or materially deprived individuals and those diagnosed with substance use disorders. Further details on selection criteria and sample characteristics have been published elsewhere [5,35]. At the fourth data collection period, the mean age of participants (N = 1871) was 44.72 years (SD: 13.86); 60% were female, and a small majority lived alone (52%).

Conceptual Framework, Variables, and Instruments
The dependent variable, "number of types of healthcare professionals consulted for MH reasons", was measured for each participant at T4. Professional consultation was defined as contact with any healthcare professional for MH reasons in the 12 months prior to the interview: General practitioner, psychiatrist, psychologist, nurse, social worker, etc. Independent variables (measured at T3) were integrated based on the literature related to predictors of MH service use [21,22,38] and were classified according to the Andersen model. Predisposing variables included: Age, gender, education, civil status, household size, number of children in the household, and health beliefs such as self-perception of mental health and self-perception of physical health, as well as satisfaction with health services. Enabling variables included: Having a family physician, previous use of MH services (consultations with any healthcare professionals or community service organizations such as self-help groups), and employment status. As MH service use is known to have a strong influence on outcomes such as quality of life, emotional well-being, and personal well-being [21], these variables were also included as enabling factors. Needs factors were categorized as: (a) Clinical needs, i.e., diagnoses (major depressive disorder, generalized anxiety disorder, post-traumatic stress disorder (PTSD), drug and alcohol dependence, as well as high psychological distress, physical illnesses, suicide ideation and (b) other health-related needs, including stressful events, unmet need for help, physical aggression, and functional disability. MH diagnoses with low prevalence such as schizophrenia or personality disorders were not included in the study, as is often the case in population-based studies [39,40]. The analytical framework is presented in Figure 2. Table 1 displays the standardized instruments used to measure the variables or dimensions identified above, including sub-dimensions measured and numbers of items, ranges for each score, interpretation of the score, and where possible the Cronbach alpha. These standardized instruments were selected as most of them were used previously in the Canadian Health Survey-Mental health and well-being (CCHS) 1.2 [41], which is however not a longitudinal population study. The dependent variable, "number of types of healthcare professionals consulted for MH reasons", was measured for each participant at T4. Professional consultation was defined as contact with any healthcare professional for MH reasons in the 12 months prior to the interview: General practitioner, psychiatrist, psychologist, nurse, social worker, etc. Independent variables (measured at T3) were integrated based on the literature related to predictors of MH service use [21,22,38] and were classified according to the Andersen model. Predisposing variables included: Age, gender, education, civil status, household size, number of children in the household, and health beliefs such as self-perception of mental health and self-perception of physical health, as well as satisfaction with health services. Enabling variables included: Having a family physician, previous use of MH services (consultations with any healthcare professionals or community service organizations such as self-help groups), and employment status. As MH service use is known to have a strong influence on outcomes such as quality of life, emotional well-being, and personal well-being [21], these variables were also included as enabling factors. Needs factors were categorized as: (a) Clinical needs, i.e., diagnoses (major depressive disorder, generalized anxiety disorder, post-traumatic stress disorder (PTSD), drug and alcohol dependence, as well as high psychological distress, physical illnesses, suicide ideation and (b) other health-related needs, including stressful events, unmet need for help, physical aggression, and functional disability. MH diagnoses with low prevalence such as schizophrenia or personality disorders were not included in the study, as is often the case in population-based studies [39,40]. The analytical framework is presented in Figure 2. Table 1 displays the standardized instruments used to measure the variables or dimensions identified above, including sub-dimensions measured and numbers of items, ranges for each score, interpretation of the score, and where possible the Cronbach alpha. These standardized instruments were selected as most of them were used previously in the Canadian Health Survey-Mental health and well-being (CCHS) 1.2 [41], which is however not a longitudinal population study.

Analyses
Univariate, bivariate, and multivariate analyses were performed. Missing values were also treated by imputation technics (however, fewer than 5% of variables had missing values). Univariate analyses consisted of frequency distributions for categorical variables and central tendency measures (mean values and standard deviations) for continuous variables. A bivariate linear regression analysis was conducted for each independent variable to determine which ones were associated with a higher number of different types of healthcare professionals consulted for MH reasons. Significant independent variables in the bivariate analyses (p < 0.10) were used to build the hierarchical linear regression model, with the alpha value set at p < 0.05 to determine which factors in the model were significant. Following the study hypothesis, clinical needs variables were entered into the model first, followed by other needs variables, then predisposing and enabling factors. Needs factors were divided into "clinical" needs factors (e.g., diagnoses) and "other" needs factors (e.g., functional disability) to highlight the differences between them in terms of impact on service utilization. For each block of variables, the goodness-of-fit was determined with the Hosmer-Lemeshow test and the variance explained using the Nagelkerke R 2 .

Results
Out of the large epidemiological catchment study, 1871 participants were selected from T4 for the present study. Within this group, 746 had either a diagnosed MD (N = 201; 27%) or high psychological distress (N = 713; 96%). Table 2 shows the main characteristics of those participants at T3, as well as bivariate associations with the number of healthcare professionals visited at T4. Regarding predisposing factors, mean age was 44 years (Table 2). Participants consisted of nearly twice as many women as men (62% vs. 38%). Most lived alone (63%). Their self-perception of physical health (2.8 on 5; SD: 1.1) and of MH (2.9 on 5; SD: 1.0) were average, as well as their satisfaction with MH services (51.7 on 100; SD: 13.8). Regarding enabling factors, over two thirds (69%) had a family physician. Over the previous 12 months, 29% (N = 219) had visited a healthcare professional for MH reasons. Of the total sample, participants most frequently consulted general practitioners (15%, N = 112), followed by psychologists at 14% (N = 104) and psychiatrists at 8% (N = 60). Sixty-two percent of those who visited a healthcare professional (N = 136/219) consulted a single healthcare professional, usually a general practitioner (25%, N = 34/136); while 26% (N = 56/219) consulted two healthcare professionals, usually a general practitioner and a psychologist (57%, N = 32/56), and 11% (N = 25/217) at least three healthcare professionals, usually a general practitioner, psychologist, and psychiatrist. Only 19% had held a job during the previous year. Regarding needs factors, major depression was most prevalent (17%), and 9% experienced suicide ideation. The psychological distress score was relatively high (13.8; SD: 5.3). The mean number of stressful events per participant was 3.8 (SD: 2.4). Physical illnesses were also prevalent among participants (64%). Independent variables significantly associated with the dependent variable, number of types of healthcare professionals consulted for MH reasons, in the bivariate analyses (Table 2) were used to build the hierarchical linear regression model. The hierarchical linear regression model is presented in Table 3 and includes four blocks: Clinical needs, other health-related needs (needs factors), predisposing, and enabling factors. The first factor, clinical needs, yielded five positively associated variables: Psychological distress, major depressive disorder, generalized anxiety disorder, PTSD, and suicide ideation. These variables remained significantly associated with the dependent variable after the introduction of the second block, other health-related needs factors, with the exception of psychological distress, and two new variables were added: Stressful events and functional disability, which were both positively associated with the dependent variable. After introduction of the third block, predisposing factors, one variable, self-perception of MH, was negatively associated with the dependent variable, meaning that those who perceived their MH as excellent were less likely to visit a greater variety of healthcare professionals for MH reasons. Regarding the fourth block, enabling factors, three variables, previous use of services for MH reasons, having a family physician, and employment status were positively associated with the dependent variable, in addition to the four previously entered variables: PTSD, stressful events, self-perception of MH, and (marginally) suicide ideation. The total variance explained by the four model was 21%: 11% for clinical needs and 2% for other health-related needs, 2% for predisposing factors, and 6% for enabling factors, respectively to the final model. The goodness-of-fit was acceptable.

Discussion
Based on this catchment area study, slightly less than one-third of the sample reported consulting healthcare professionals for MH reasons in the previous year. This result falls within the low range of consultations (from 5% in low income to 44% in high income countries) observed in previous population studies [22,55,56], which may be explained by the inclusion of individuals with high psychological distress in our sample of diagnosed individuals with MD. Among those who consulted a single healthcare professional, general practitioners and psychologists were equally favored, followed by psychiatrists. This order in the choice of healthcare professionals is consistent with previous Quebec epidemiological studies [24,57]. The finding that fewer than four in ten individuals had used at least two healthcare professionals makes sense, as psychotherapy and treatment with medication are generally recommended in combination with the management of MH conditions in step-care models [16,58]. The finding that a minority (about 10%) required consultations with psychiatrists corresponded to the rate of more complex, recurrent, or serious MD in the sample [59]. It is well known that general practitioners have difficulty treating such complex cases in primary care settings [60].
The hierarchical regression model revealed that needs factors predicted the highest number of types of healthcare professionals consulted for MH reasons, followed by enabling factors; whereas the contribution of predisposing factors was marginal. These findings confirm our hypothesis that individuals with complex or serious needs would be more inclined to consult a greater number of different types of healthcare professionals. However, it should be noted that there were equivalent numbers of significant needs factors (PTSD, stressful events, and (marginally) suicide ideation) and enabling factors (previous use of MH services, having a family physician, and employment status) in the final model; whereas only one predisposing factor (self-perception of MH) remained significant. Interestingly, three needs factors (major depressive disorder, generalized anxiety disorder, and functional disability) ceased to be associated with the dependent variable after the block of enabling factors was introduced into the model, suggesting that the presence of enabling factors substantially reduced the influence of these needs factors on number of different types of healthcare professionals consulted for MH reasons.
As predicted in this study, and reported previously [24,61], we found that the presence of a diagnosed MD was associated with the likelihood of consulting a greater number of healthcare professionals. Specifically, two clinical needs variables (PTSD and (marginally) suicide ideation) were associated with the dependent variable in the final model. A Canadian study found that individuals with PTSD usually experience chronic symptoms causing functional impairment and high rates of comorbidity [62]. Another study found an association between PTSD and substantial reduction in quality of life affecting general health, energy levels, emotional and physical well-being, as well as social functioning [63]. Reports indicate that individuals with suicide ideation are also more likely to seek treatment from both a general practitioner and MH specialist [64]. This may reflect a perceived need for MH care on the part of individuals with suicidal thoughts and behaviors [65]. Among individuals with suicide ideation, perceived need for MH care has been associated with increased likelihood of service use [65]. A study on MH care for adults with suicide ideation [66] found that perceived need for care predicted nearly four times the likelihood of receiving treatment. 'Stressful life events' was the only other health-related needs variable associated with the number of types of healthcare professionals consulted for MH reasons. There is considerable evidence that individuals who have experienced an above average number of negative life events are more likely to experience psychological distress [67]; they also have greater odds of developing PTSD [68] or other MD [69][70][71] and are more likely to seek MH services [68,72].
Three enabling factors predicted consultations with a greater number of different types of healthcare professionals for MH reasons: Previous use of MH services, employment status, and having a family physician. Due to the chronic nature of certain MH conditions, episodes of relapse or recurrence [4] may require adherence to treatment over a longer period for achieving better outcomes. Past research has also found that prior treatment history for MD was associated with more positive attitudes toward medical interventions [73] and help-seeking from various providers [74]. Knowledge and experience with services are recognized as facilitators in accessing care [75]. Having a family physician also provides an entry point to the healthcare system: Family physicians diagnose and treat patients who present with MH symptoms, making referrals of more severe or complex cases to specialized MH professionals [76,77]. By contrast, not having a family physician may result in unmet needs for care [78] or delayed access to MH care, especially among individuals unaware of their MH symptoms and the available resources [75]. This finding underscores the importance of integrating MH into primary care, improving continuity of care by making family physicians universally available, and increasing their training in the screening and treatment of MH problems, and better supporting them through collaborative care models. Employment status was another predictor of the number of different types of healthcare professionals consulted for MH reasons. Quite possibly, employed individuals have greater access to certain healthcare professionals, including private psychologists [5]. Moreover, employee assistance programs may offer workplace educational support, counseling, confidential screening, and assessment [79]. While unemployed individuals may feel depressed or have a greater need for MH care [80], those who occupy positions with high levels of job insecurity or negative stress may be at even higher risk of MH problems. For instance, one study [81] reported that employed individuals with perceived job insecurity were more likely to have poor MH status and greater need for care than unemployed individuals; they also made greater use of MH services [81].
Finally, one predisposing factor, self-perception of MH, negatively predicted the number of different types of healthcare professionals consulted for MH reasons. This corresponds to previous findings suggesting that patients who had a worse self-perception of their MH were more likely to seek help from general practitioners and MH specialists rather than relying on a single healthcare professional [82,83]. Another study [24] found that individuals who rated their MH worse were more likely to seek help from either a general practitioner and psychologist, or a general practitioner and psychiatrist. This finding underscores the relevance of individual perceptions on the need for care in the absence of a MD and MH service utilization.

Limitations
The findings of this study should be considered in light of certain limitations. First, the frequency of consultations with each healthcare professional was not taken into account. As such, continuity and appropriateness of care were not considered, which are key dimensions impacting quality of care and patient recovery. Second, the diversity of types of healthcare professionals consulted did not take into account the existence of actual collaboration among them, as required in collaborative care models. Third, some pertinent predictors of consultations with different types of healthcare professionals for MH reasons that may influence service utilization, including lifetime disorders, symptoms severity, suicide attempt [27], ethnicity, or religion [84] were not measured in this study. Fourth, populations younger than 15 and older than 65 were excluded from the study, as MH health service use patterns for these age groups are very distinct. Fifth, the study did not cover the full spectrum of MD, as some disorders assessed in the third data collection period were not assessed in the fourth data collection. Serious MD with very low population prevalence, such as schizophrenia and personality disorders, were also excluded. Six, the final study sample (at T4) was less representative of the population (i.e., in terms of youth, individuals with high material deprivation, and substance use disorders). Finally, the total variance (21%) explained by our model was relatively low.

Conclusions
This study was innovative in assessing predictors of diversity in the number of types of healthcare professional consultations for MH reasons among individuals, not only those diagnosed with MD, but mainly those affected by high psychological distress, using the Andersen Behavioral Model. This latter clientele is often overlooked by MH professionals, who focus mainly on diagnosed MD cases in their practices. Findings revealed that needs factors were most strongly associated with the number of types of healthcare professionals consulted, but that enabling factors may reduce the influence of certain needs factors, particularly those involving clinical variables. Multiple variables associated with consultations of a greater number of different types of healthcare professionals for MH reasons were identified: PTSD, stressful events, previous use of MH services, employment status, having a family physician, self-perception of MH, and (marginally) suicide ideation. The identification of specific predictors may help orient MH programs and interventions. Considering that enabling factors are easier to tackle than needs or predisposing factors, it would be important for MH managers to prioritize the development of collaborative care models to facilitate continuity and appropriateness of services for individuals who view their MH as poor. Moreover, all individuals with MH problems should have a family physician to provide an entry point to the healthcare system. Family physicians are responsible for addressing a wide range of MH needs, but also for making referrals to specialized MH services in more severe or complex cases, such as PTSD. Further extensions of employment assistance programs may facilitate better MH, as such support in the workplace may serve as an antidote to stress and related MH problems, while providing individuals with the means to consult MH specialists when needed. Finally, information on help-line services and crisis centers should be more widely disseminated, given the numbers of individuals affected by stressful events and suicide ideation.