Clinical depression is one of the most common chronic disorders managed by general practitioners, accounting for 4.3% of all encounters (1). Studies of general practices in Australia have highlighted the extent to which these disorders are not only common but also chronic and disabling (2). Over the past decade there has been a large expansion of secondary care services for people with depression, primarily time-limited psychological interventions by specialist professionals (3–5).

The promotion—hence direct financing—of shared-care models has been the one constant theme of mental health reform in primary care (6,7). However, it remains unclear whether access to mental health specialists is determined by clinical need or by sociodemographic characteristics. The rational view would be that persons whose treatment is managed through shared care have more severe, chronic, or disabling illnesses and are in greater need of care. An alternative view is that socioeconomic, geographic, or other factors drive such access. To date, Australian studies of access and equity have used routinely collected data and have produced conflicting results (4,5). An alternative method is to examine how physicians proceed once they have identified clinical depression and what influences management and referral.

In this study we aimed to evaluate the clinical management for people identified by their general practitioners as clinically depressed. We also sought to determine whether clinical need or sociodemographic characteristics drove access to mental health specialists.

In October 2009, all 21,282 general practitioners in Australia listed in a database maintained by Dendrite Clinical Systems in Sydney were contacted by mail inviting expressions of interest (EOIs). From these EOIs a random sample of 200 general practitioners was selected and stratified by state and by metropolitan versus rural area to ensure that the sample of participants reflected the geographical distribution of the Australian adult population. If recruitment targets for a geographic stratum were not met, stratum-specific EOIs were remailed. If a selected general practitioner withdrew, the process defaulted to the geographically closest general practitioner having expressed interest. To achieve a geographically balanced sample of 200 participants, it was necessary to contact an additional 200 general practitioners.

Each general practitioner prospectively aimed to recruit five to seven consecutively presenting patients. The inclusion criteria were age 18 years or over and a clinical diagnosis of depression made by the doctor, irrespective of the reason for that specific consultation. In several practices, results from a Somatic and Psychological HEalth REport (SPHERE-12) (8), routinely completed by patients prior to the consultation, could be used by the general practitioner to aid diagnosis. Patients were excluded if their doctor considered their depression to be due to an acute life event or to bereavement.

The study was approved by the Royal Australian College of General Practitioners National Research Evaluation Ethics Committee, with all participating doctors and patients providing signed, informed consent.

General practitioners completed a case report form containing details of this depressive episode, including primary diagnosis, number of previous episodes, age at onset, duration of episode, symptoms and their severity, medication prescribed and indication, and any other additional interventions, including specialist care, indicated in the clinical record. Patients completed a questionnaire that included the nine-item depression scale of the Patient Health Questionnaire (PHQ-9) (9) to evaluate the presence of a major depressive syndrome and the severity of the depressive symptoms with a standard scoring protocol, the Pittsburgh Sleep Quality Index (10), and questions assessing sociodemographic characteristics, including age, gender, ethnicity (Caucasian versus other), language spoken at home (English versus other), marital status (never married versus married or partnered versus widowed or divorced), education level (primary versus secondary versus tertiary), employment and benefit status, and disability days for this episode (11).

Participants whose doctor recorded a primary psychiatric diagnosis other than depression (with or without anxiety) or who did not report data were excluded from the analyses. Descriptive statistics for the whole sample were calculated, and t tests for continuous data and chi square tests for categorical data were used to compare differences between participants whose treatment was managed solely in primary care and participants in shared care.

A total of 839 participants were recruited (range one to seven per doctor) between November 2009 and September 2010. Of these, 754 (90%) had a primary psychiatric diagnosis of clinical depression made by the general practitioner. The remaining 85 participants had another primary psychiatric diagnosis, predominantly bipolar disorder (N=54). Nineteen (3%) had no patient self-report data, leaving a final sample of 735 participants with both physician- and self-report data. Data completion was missing for individual variables for between zero and 160 participants. For 309 (42%) participants, care during this episode was shared by a psychiatrist, psychologist, or counselor. There were no differences in the proportions of patients referred to shared care by metropolitan (N=183 of 416, 44%) and rural (N=93 of 231, 40%) doctors.

As shown in Table 1, participants were predominantly female (65%), middle-aged (mean=49.4 years; range 17–89 years), and Caucasian (90%), and 97% spoke English at home. Of those aged 18 to 65 years (N=598), only 47% worked either full- or part-time, and 32% received some form of income compensation—the Disability Support Pension was the most common (15%).

There were no demographic differences between participants in primary care alone and those in shared care, except for age. Older people were less likely to be referred for shared care (mean difference=2.3 years; 95% confidence interval [CI]=.03–4.60). Participants in shared care, however, were more likely than participants who received care only from a general practitioner to receive some form of income support (39% and 27%, respectively).

Participants’ depression demonstrated a high level of recurrence and chronicity—54% of participants had at least two previous episodes and for 47%, the current episode had lasted for over 12 months. Of the 700 participants with adequate PHQ-9 data, 55% (N=384) met criteria for a current major depressive syndrome (at least one cardinal symptom and a total of five of nine symptoms reported as being present more than half the days) (9). General practitioners identified moderate or more severe depression among 355 of the 424 (86%) patients who reported depression of moderate or greater severity. Significant sleep disturbance was extremely common (86%). Participants in shared care had more chronic, recurrent conditions and more severe symptoms on all measures and reported more than twice the number of disability days than participants who received care only from a general practitioner (20.0±56.4 versus 9.7±33.9 days, mean difference=10.3, CI=2.2–18.3).

Prescription use was common, with 77% (N=566) taking an antidepressant and 30% (N=222) an anxiolytic or hypnotic. Chronic medication use was highly prevalent. Use of the current medication for over 12 months was identified among 212 of 526 (40%) users of antidepressants, 58 of 110 (53%) users of anxiolytics, and 40 of 109 (37%) users of hypnotics. Patients under shared care were more likely to be taking an antidepressant longer—111 of the 237 (47%) participants referred for shared care who had been taking antidepressants had done so for over 12 months compared with 101 of 289 (35%) participants receiving care from a general practitioner only (odds ratio [OR]=1.64, CI=1.15–2.33). Conversely, among participants taking hypnotics, those in primary care were much more likely (30 of 67 participants, 45%) than those in shared care (ten of 42 participants, 24%) to have taken them for over 12 months (OR=2.60, CI=1.10–6.12). Similar numbers of people took a second-generation antipsychotic, mainly quetiapine or olanzapine, or a standard mood stabilizer, such as lithium (N=36 and 40, respectively, of 734 participants, each 5%).

This snapshot of primary care management for depression suggests that for most individuals, their depression was significantly disabling. Recurrent, chronic conditions with high levels of both doctor- and patient-rated symptoms, receipt of income support and disability pensions, and chronic medication use were common, even in the group managed solely in primary care. Over half of this sample had a current major depressive syndrome according to an external measure, the PHQ-9.

These findings contrast with studies that screen for depression in primary care settings, where “transient” and milder forms of depression may be more common. For instance, a recent Australian primary care study (12), which identified people with depressive symptoms through screening, found that only 26% satisfied the same PHQ-9–derived depression criteria. This does not necessarily reflect comparative overidentification of depression in this study, given that the patients could be in any stage of their current depressive episode, including recovering, so their symptoms might not be as severe as a full syndrome at the time of assessment. Supporting this, the general practitioners correctly identified the level of severity of 86% of patients reporting moderate or greater symptoms.

About one in eight (12%) of all primary care contacts for depression in Australia involve a referral to an allied health professional, and referrals to psychiatrists occur in management of 1.8 per 100 mental health–related problems (1). For a significant minority (42%) of participants in this study, care was shared with a specialist, indicating that the sample was likely biased toward a more impaired population. Although those in shared care were more severely affected on all clinical measures, we could identify no other sociodemographic factor, apart from older age, that was associated with accessing specialist care. If anything, the higher rates of income support and receipt of disability pensions in the shared-care group would indicate poorer socioeconomic status. Conversely, there was a trend for those in shared care to be more educated. Notably, the proportion of patients in shared care among rural and urban general practitioners was similar. Although it is likely that a number of other nonclinical factors determine who accesses shared care, such as a lack of partnership formation and clinician attributes (13), this study demonstrated that clinical factors are a major driver.

The percentage of participants in this study (77%) who used antidepressants was similar to other reports about antidepressant use in Australian primary care. In 2008, an almost identical percentage (76%) of depressed patients in general practice in Australia took an antidepressant (1). Data on anxiolytic and hypnotic use are less readily available, although anxiolytics and hypnotics together comprise 4.5% of all general practitioner prescriptions (1). Although the duration of antidepressant use is understandable given the chronic or recurrent conditions of the participants of this study, the long duration of anxiolytic and hypnotic use is questionable (14). The significant number of participants who used anxiolytics for at least 12 months underscores the difficulties experienced by both clinicians and patients in trying to reduce benzodiazepine usage (15). The lower rate of extended hypnotic use among participants in shared care suggests that shared care may be more successful in reducing the need for coprescribing, either through the use of nonpharmacological methods or new treatment strategies.

With nearly nine of ten patients reporting clinically significant sleep disturbance, strategies focused on addressing sleep and circadian dysfunction would seem pressing. The use of second-generation neuroleptics to manage depression of one in 20 patients is potentially notable. Given the cardiometabolic side effect profile of these drugs, the trend and impact of such prescribing for depression, now approved by the Therapeutic Goods Administration, is worth monitoring.

Because the sample was consecutive patients identified by their general practitioner as undergoing a current depressive episode, this study had limitations. It was likely to be biased toward those with more severe conditions or those needing medication review and to underrepresent marginalized groups who consult less frequently. A comparison with routine primary care data (1) suggests that these selection and generalizability issues may be limited; for example, new cases account for 16% of contacts for depression in routine primary care compared with 11% of contacts for depression in the first month of this study; in addition, the mean age of depressed patients identified through routine primary care monitoring (1) and those in this study was similar. Finally prescription levels in routine primary care were almost identical to those observed here.

The questions were derived from well-validated instruments and national surveys. The doctors and patients were blinded to the study’s hypothesis and thus were unlikely to systematically record data in a biased way.

Depressed patients identified by Australian general practitioners exhibited chronic, recurrent, and moderate to severe disabling conditions, with many taking antidepressants and other medications over prolonged periods. Chronic usage of hypnotics and anxiolytics and the emergence of second-generation antipsychotics in depression management may require closer evaluation. General practitioners triaged and referred more severely affected patients to other health professionals for shared care, and sociodemographic characteristics seemed related to little, if any, inequity in access.

This study was performed in collaboration with Servier Laboratories Australia, which provided financial support. The National Health and Medical Research Council Australia also provided financial support through a fellowship (464914) to Dr. Hickie.

Research conducted at the Brain and Mind Research Institute Mental Health Program is supported by Servier. Dr. Glozier has received travel expenses from Servier to attend conferences presenting this study's findings and an honorarium from Servier for an expert statement on depression and sleep. Dr. Hickie has received travel support, research support, or educational support from Servier, AstraZeneca, Pricewaterhouse Coopers, Pfizer, and Eli Lilly and Company. Ms. Davenport reports no competing interests.