Five South Australian Divisions of General Practice were contacted in February 2004, and a letter inviting GPs to participate was sent to about 400 GP members. A total of 45 GPs from 23 urban and rural practices responded and agreed to participate.

GP recruitment of patients began in May 2004, and the study ended in December 2005. GPs identified patients aged 18 years or older who met diagnostic criteria for a depressive disorder according to the Diagnostic and statistical manual of mental disorders, fourth edition (DSM-IV).1 Other patient inclusion criteria were availability to be followed up for the next 12 months, and ability to give informed consent. Patients were excluded if they were known to be undergoing a separate treatment program, were suffering from psychoses, or were unable to complete the English-language questionnaires or interview.

Sample size calculations indicated that 120 patients would be required in each of the intervention and control groups (95% confidence, 80% power, assuming a design effect of 1.5 due to clustering).10

KBA is based on the literature, current thinking and clinical experience.8 Medication is used as clinically indicated, and the program is started once the patient’s depression has been stabilised by initial treatment. KBA involves a novel, multimodal, skills-based approach. It incorporates 10 steps and utilises a range of evidence-based psychosocial strategies, such as problem solving (Box 1).12,13

Intervention-group GPs and patients were provided with resources including a GP training manual or patient manual and a purpose-designed relaxation CD.14,15 GPs in the intervention group completed 20 hours of training, conducted by a GP and a psychologist. Their training kit contained information on depression, the study protocol, assessment tools, and skills training related to the KBA program. GPs in the control group completed a 3-hour training session on the study protocol.

To evaluate the impact of the intervention on depression relapse rates, our analysis looked at overall relapse rather than at individual time points. We used χ2 analyses to determine the relationship between relapse and identified variables (study group, sex, age, duration of symptoms, past history of depression), and calculated relative risks (RRs). For the variable of age, a logistic regression was performed. Finally, we used a model that included all the main effects and each variable’s interaction group. Interactions with a P value less than 0.20 were then included. In the final regression model, the variables of age, sex, group and age/group were retained.

As there was variability in when the DASS and the WHOQOL-BREF scales were administered, the results were analysed according to days since baseline. A mixed-models approach was used due to some missing data. The variables group, sex, age, history of depression, length of time with symptoms, antidepressant use, and location were used to build a regression model to describe changes in DASS depression scores over time. The impact of KBA on quality of life was measured across physical, psychological, social and environmental domains of the WHOQOL-BREF, and results were examined using a mixed-model regression analysis.

Data were analysed using SAS, version 9.1 (SAS Institute, Cary, NC, USA).

Ethics approval was received from the Human Research Ethics Committee at the University of Adelaide.

There were no statistically significant differences between the two groups in depression relapse rates (χ21 = 1.51; P = 0.23). However, there was a non-significant tendency for relapse to be reduced in the intervention group (RR = 0.77; 95% CI, 0.50–2.05) (Box 4).

In the final regression model, age was predictive of depression relapse, with increased age associated with decreased risk of relapse. Older patients (≥ 50 years) in the intervention group showed a significantly lower probability of relapse than those in the control group (P = 0.018) (Box 5). “Older” refers to one standard deviation above the mean age (where mean = 39.24 and SD = 11.30). Younger people in either group tended to show similar probability of relapse.

Sex approached significance, with women in both groups less likely to relapse (P = 0.069).

Five GPs and eight patients were interviewed to ascertain their perceptions of the KBA program. Interview results indicated that the content of KBA was viewed by GPs and patients as highly or extremely relevant. The most helpful steps identified by GPs included Step 2 (information on depression), Step 4 (coping skills) and Step 5 (cognitive behavioural therapy). All of the GPs felt the KBA program provided the right amount of information and said that they were highly or extremely satisfied with it.

The most helpful steps identified by patients included Step 1 (assessment and goal-setting), Step 4 and Step 5. Seven of the eight patients interviewed indicated the KBA program provided the right amount of information. Two patients were satisfied and six were highly or extremely satisfied.

One GP commented that, “Most people with anxiety and depression can benefit from this type of program”. A patient stated, “‘Keeping the blues away’ helped me. It opened my eyes”, while another noted, “It is supportive and enlightening. You can take what is needed at the time. It is a well planned program with something for everyone”.