An Expert System to Reduce Depression in Primary Care

Two statistical criteria (Jacobson \& Truax, 1991a; Atkins, Bedics, McGlinchey, \& Beauchaine, 2005) were used to define reliable and clinically significant improvement. The first involved selecting a cutoff that represents remission or the absence of symptoms, which was selected to be BDI-II < 9. The second involved selecting a pre-post difference score that represents a statistically reliable change (e.g., how much change-1 point, 5 points, 10 points-is needed to be 95% confident that a real change has occurred, rather than just a chance fluctuation due to the unreliability of the measure?) Using a general formula for calculating reliable change that incorporates test-retest reliability (Jacobson \& Truax, 1991b), reliable change for the BDI-II was calculated to be 5.13, and rounded down to 5. Thus, reliable and clinically significant improvement on the BDI-II was defined as a reduction of 5 or more points from baseline to follow-up and a follow-up score < 9.

Depression prevention was defined as: "Using effective methods to keep depression from occurring, or if it does occur, to keep it as mild and brief as possible." Effective methods were: 1) controlling negative thinking; 2) engaging in healthy, pleasant activities; 3) practicing stress management; 4) exercising; and 5) getting professional help when needed. Patients who reported that they were not currently practicing depression prevention and had no intention of doing so in the next 6 months were classified in the precontemplation stage; those who intended to practice depression prevention in the next 6 months or next 30 days were classified in the contemplation or preparation stage, respectively; those who had been practicing depression prevention for less than 6 months were in the action stage, and those who had been practicing for more than 6 months were in maintenance. This outcome represents the number of participants in the action or maintenance stage at 9 months follow-up.

At baseline and follow-up, Major Depression was assessed using 9-item depression module of the Primary Care Evaluation of Mental Disorders Patient Health Questionnaire-PHQ-9 (Spitzer, Kroenke, \& Williams, 1999). In this analysis, Major Depression was assessed among participants not meeting criteria for major depression at baseline.

At 9 months follow-up, participants who had been prescribed antidepressant medication were asked if they had started and were still taking it. Participants who were taking their medication or had stopped with their doctor's advice were considered to be adherent.

Physical functioning was assessed using the Physical Functioning, Role Functioning, Health Perceptions, and Pain subscales of the 20-item Medical Outcomes Study Short Form survey (SF-20) (Stewart, Hays, \& Ware, Jr., 1988). (SF-20 subscales assessing mental health and social functioning were omitted to get a purer measure of physical functioning). Physical functioning was computed by taking the mean of the four subscales after each was linearly transformed to range from 0-100. The change scores reported here are the difference between the physical functioning scores at baseline and 9 months (i.e., 9 months minus baseline). Change scores range from -100 to +100, with higher positive scores indicating more improvement in physical functioning.