This study, carried out in the context of a collaborative care program for common mental disorders, is aimed at identifying the predictors of Primary Care Physician (PCP) referral to Community Mental Health Center (CMHC) and patterns of care. Patients with depression or anxiety disorders who had a first contact with CMHCs between January 1, 2007–December 31, 2009 were extracted from Bologna Local Health Authority database. A classification and regression tree procedure was used to determine which combination of demographic and diagnostic variables best distinguished patients referred by PCPs and to identify predictors of patterns of care (consultation, shared care, and treatment at the CMHC) for patients referred by PCPs. Of the 8570 patients, 57.4% were referred by PCPs. Those less likely to be referred by PCPs were living in the urban area, suffered from depressive disorder, and were young. As to the pattern of care, patients living in the urban area were more likely to receive shared care compared with those living in the nonurban area, while the reverse was true for consultation. Predictors of CMHC treatment were depression and young age. Prospective studies are needed to assess length, quantity, and quality of collaborative treatment for common mental disorder delivered at any step of care.
Mental disorders are very common in primary care setting: the WHO Collaborative Study on Psychological Problems in General Health Care (PPGHC) reported a global prevalence of 24.0% [
In Italy, primary care is placed at the heart of the health care system. Primary care physicians (PCPs) are independent contracted professionals who operate under the control of Local Health Authorities (LHAs).
On average, LHAs are responsible for the overall health of, and for the services offered to, a target population of 350,000 inhabitants. PCP are the first contact for the most common health problems and act as gatekeepers for drug prescription and for access to specialty and hospital care.
Specifically, PCP are involved in delivering various primary care services like health promotion and preventive care activities, diagnosis, treatment, and followup of noncomplex, acute, and chronic conditions. They also have an increasing role as coordinators of services provided to patients with chronic diseases [
Mental health care in Italy is currently delivered by Mental Health Departments (MHDs) that are in charge of the management and planning of all medical and social activities related to prevention, treatment, and rehabilitation in a defined catchment area. Within the departments, Community Mental Health Centers (CMHCs) cover all activities pertaining to adult psychiatry in outpatient settings and manage therapeutic and rehabilitation activities delivered by day care services and nonhospital residential facilities [
Among the existing attempts to promote the integration between primary care and mental health services, the “G. Leggieri” Program was started in 2000 as an effort of the Health Government of Emilia-Romagna Region to coordinate initiatives of primary care-mental health cooperation undertaken since the 1980’s. A steering group of representatives of local health authorities, mental health and primary care departments, scientific associations, and academic institutions was established in order to deliver specific recommendations about the organisation of collaborative activities. In particular, two main objectives were identified and pursued: (1) to improve the quality of treatment for patients with common psychiatric disorders in primary care; (2) to modify the pathways of care, supporting the management of common psychiatric disorders in primary care and focusing mental health services’ activities towards severe or difficult-to-treat cases.
In the framework of this program, regional recommendations were delivered according to the stepped care model proposed in the NICE guidelines for depression and anxiety [
Within this regional context, local models of collaborative care have been gradually refined [
The specific aims of present study are (1) to determine in which proportion patients with common mental disorders who have a first contact with the CMHCs are referred by PCPs; (2) to identify the predictors of PCP referral versus other referral sources; (3) to predict the pattern of care of patients referred by PCPs, as a function of their diagnosis and demographic characteristics.
The data source for this study is the mental health information system of the Bologna LHA. The Bologna LHA is one of the largest in the country and serves approximately 850,000 inhabitants, roughly one fifth of the regional population.
The mental health information system was implemented in 2007 for administrative and clinical-epidemiological purposes. All patients who had at least one contact with community-based mental health services were recorded in the database since then, reflecting the total secondary mental health care in the area. Data include, in addition to patient’s ID number, demographic characteristics, the patient’s PCP name, the ICD-9 CM diagnosis, information on each type of intervention administered, and the number and type of staff involved in the intervention.
Demographic and diagnostic information of patients who had their first contact with one of the 11 CMHCs of the LHA between January 2007 and December 2009 was extracted from the database. Patients who were not living in Bologna province, those with a missing diagnosis or with fewer than 18 years of age were excluded from the analyses.
ICD-9 CM diagnoses were classified into 9 groups: schizophrenia (295, 297, 298 excl. 298.0, 299), depression (296.2-3, 296.9, 298.0, 300.4, 309.0, 309.1, 311), bipolar disorders (296.0, 296.1, 296.4–8), personality disorders (301, 302, 312), alcohol and substance use disorders (291, 292, 303, 304, 305), dementia and organic mental disorders (290, 293, 294, 310), anxiety and somatoform disorders (300 (excl. 300.4), 306, 307.4, 307.8-307.9, 308, 316), mental retardation (317, 318, 319), and other mental disorders (307.0–307.3, 307.5–307.7, 309.2–309.9, 313, 314, 315).
Patients with depression or anxiety and somatoform disorders, who are the target of Leggieri Program, were retained for the classification tree analyses.
As mentioned above, there are 3 steps that entail direct CMHC involvement after referral by PCP.
This includes the establishment of a psychiatric diagnosis, significant life events, and the description of possible dysfunctional coping behaviors. The evaluation is followed by suggestions for the treatment plan, which is delivered by the PCP. Information is then forwarded to the PCP in a typed report designed to be thorough, but concise.
After the assessment, the consultant provides brief and focused therapeutic interventions to support the PCP management of psychiatric disorders. Written communications are accompanied by telephone communication or interpersonal contacts in order to increase understanding and cooperation between psychiatrists and PCPs. For example, the psychiatrist could start pharmacological treatment and furthermore evaluate the initial treatment response and patient compliance. In other cases, a brief psychological intervention can be provided (counseling).
Severe and complex cases are referred to the CMHC that takes full responsibility for psychiatric management.
In this study, patients referred to the CMHCs of Bologna area with a diagnosis of anxiety or depressive disorder were classified into two mutually exclusive groups according to the referral source (PCPs versus other sources).
To determine which combination of demographic and diagnostic variables best distinguished patients referred by primary care physicians from those referred by other sources, data were analyzed using a classification and regression tree (CART) procedure.
This procedure was also used to determine which characteristics (and combination of characteristics) predicted the pattern of care of patients referred by primary care physician (consultation, shared care, and treatment at the CMHC).
In contrast to traditional statistical models, CART is a nonparametric analysis that simultaneously examines interactions between continuous or categorical variables to create a decision tree. Researcher bias is limited as CART can use large numbers of variables to create a decision tree and cutoff on continuous variables that are defined by the procedures.
To date, there are no studies that employed these methods to predict outcomes such as referral to PCP or patterns of care in patients referred to CMHC, but this method has been recently used in psychiatry [
The CART procedures build decision trees beginning with a root node that includes all cases, then the tree branches into two subgroups (or nodes) and grows iteratively by identifying optimal cut points for continuous discriminating variables in the predictor set. Categories of nominal variables (such as diagnosis or marital status) are merged by the procedure if the distribution of the dependent variable is similar across the categories. The best discriminating predictor is selected first, and then subsequent predictors are entered into the procedure if they contribute significantly to subtyping cases into homogeneous groups. Variables not useful to discriminate cases do not enter into the procedure. The tree grows until a stopping criterion is met or no further significant improvement in the classification of study participants is possible. At the end of the procedure, the study population is partitioned into terminal nodes that are as homogeneous as possible with respect to the categories of the dependent variable. The final tree is “pruned” to avoid model overfitting. This is done by a procedure that, after the tree is grown in its full depth, trims it down to a smaller subtree that has an acceptable risk of misclassification (defined as 1 standard error with respect to the risk in the full tree). All analyses were carried out using SPSS, version 17.0 (Chicago, IL).
8570 patients with depression or anxiety and somatoform disorders, that constituted 56.4% of first contacts at the CMHC were included in the classification tree analysis. The proportion of patients with these disorders referred by primary care physicians to the CMHCs was 57.4%.
The proportion of PCP referrals increased from 51.5% to 62.2% in the period 2007–2009.
The classification tree analysis included as independent variables diagnosis, age, gender, educational level, marital status, nationality (Italian versus other), and area of residence (urban versus nonurban).
Results (Figure
Classification tree analysis showing the predictors of PCP referral versus other referral sources.
In summary, among patients with common mental disorders in contact with CMHCs, those less likely to be referred by PCPs were living in the city, suffered from a depressive disorder, and were young. This suggests that young urban depressed patients could face barriers in help-seeking for a variety of factors. As widely recognized [
Of the 4913 patients referred by PCPs, 1276 (26%) received a consultation, for 765 (15.6%) shared care was agreed between the CHMC and the PCP and for the remaining 2872 (58.5%) the CMHC was exclusively in charge of the intervention. A strong increase in the shared care pattern was observed from 2007 to 2009, paralleled by the decrease of treatment at the CMHC (Table
Pattern of care distribution over time in patients referred by PCPs.
Year of first contact | Total | |||||
2007 | 2008 | 2009 | ||||
Pattern | Consultation | 360 | 384 | 532 | 1276 | |
% | 24.9% | 22.3% | 30.5% | 26.0% | ||
Shared care | 52 | 281 | 432 | 765 | ||
% | 3.6% | 16.3% | 24.7% | 15.6% | ||
Treatment at CMHC | 1033 | 1056 | 783 | 2872 | ||
% | 71.5% | 61.4% | 44.8% | 58.5% | ||
Total | 1445 | 1721 | 1747 | 4913 | ||
% | 100.0% | 100.0% | 100.0% | 100.0% |
The classification tree analysis (Figure
Classification tree showing the predictors of pattern of care among patients referred by PCPs.
In summary, we observed a different pattern of collaborative care in urban compared with nonurban areas. Consistent with our expectations, predictors of CMHC treatment were depression versus anxiety disorders and young age versus older age.
Our results should be interpreted keeping in mind that the focus is on patients with common mental disorders
Patients with common mental disorders (anxiety or depression) comprised 56.4% of referrals, more than half of them (57.4%) were referred by PCPs with a trend rapidly growing over time (from 51.5% to 62.2%). The pathway from PCP to CMHC appears to have strengthened in comparison to that reported in other previous Italian investigations, such as in South-Verona, where PCP referrals accounted for 40% of new cases in 2000 [
The classification tree analysis suggests that the residence area plays an important role in the PCP referral process that appears to be more active in the nonurban compared to the urban area. In our opinion this finding is partially related to the organizational characteristics of Primary Care Services. In fact, urban PCPs often run individual practices, whilst nonurban PCPs are frequently associated in group practices, located in the same outpatient clinic. This organization fosters access, continuity of care, training opportunities and plays a part in the integration with mental health services. Moreover, nonurban practitioners have traditionally stronger links with their community and other health services (including mental health services), which favors their role of gatekeepers to secondary care [
In the urban area older patients had increasing odds to be referred to CMHCs compared with younger patients. A possible interpretation is that patients with an age of 63 years or more are probably more familiar with their PCP and more inclined to endorse their psychological complaints during the visit. Among patients <63 years of age, PCP referral was more common for anxiety than for depressive disorders. Of note, among depressed patients an age <30 years seemed to further disfavor PCP referral.
By and large, these data highlight possible barriers that young people have to face in accessing health services and their unmet needs especially in urban areas. As many authors have noted [
As to the pattern of care of patients referred by PCPs, examined in the second classification tree, we found that shared care was more frequent in urban than in nonurban areas, while consultation was more frequent in nonurban areas. Increasing either consultation or shared care was a targeted priority of the Leggieri Program, but in both areas a nearly 60% of common mental disorders referred by PCPs still receive specialized treatment at the CMHC, although this percentage was decreasing over the investigated three-year period (from 71.5% in 2007 to 44.8% in 2009). However, in the next future we expect a further decrease of this pattern of care for common mental disorders, as a result of increasing cooperation and PCP training.
It is remarkable that other demographic characteristics such as gender, marital status, educational level, and nationality, that traditionally play a role in the help-seeking behaviors and in service utilization patterns [
Overall, our findings provide a preliminary evidence on the implementation of the Leggieri Program in Bologna LHA. Future perspectives include the design of policies aimed at removing barriers in accessing mental health services for young people, mainly in the urban environment. Moreover, prospective studies using the Bologna LHA mental health database are needed to assess length, quantity, and quality of treatment delivered at any step of care in order to ensure that patients with common mental disorders receive appropriate and effective integrated care.