Cost-effectiveness of integrated care for elderly depressed patients in the PRISM-E study

One proposed strategy to improve outcomes associated with depression and other behavioral health disorders in primary care settings is to strengthen collaboration between primary care and specialty mental health care through integrated care (IC).

We compare the cost-effectiveness of IC in primary care to enhanced specialty referral (ESR) for elders with behavioral health disorders from the Primary Care Research in Substance Abuse and Mental Health study, which was a randomized trial conducted between 2000 and 2002, using a societal perspective.

The IC model had a behavioral health professional co-located in the primary care setting, and the primary care provider continued involvement in the mental health/substance abuse care of the patient. The comparison model, enhanced specialty referral (ESR), required referral to a behavioral health provider outside the primary care setting, and the behavioral health provider had primary responsibility for the mental health/substance abuse needs of the patient. Costs and clinical outcomes for 840 elders with depression were analyzed using incremental cost-effectiveness ratios, the net benefits framework, cost-effectiveness planes, and acceptability curves. Outcomes were measured by the Center for Epidemiologic Studies Depression Scale (CES-D) and converted to depression-free days and Quality Adjusted Life Years (QALY). A variation on depression free days was proposed as an improvement on current methods. Separate analyses were conducted for Veteran's Affairs (n=365; n=175 in IC and n=190 in ESR) and non-Veteran's Affairs (n=475; n=242 in IC and n=233 in ESR) settings.

ESR participants in the non-VA sample exhibited lower average CES-D scores (i.e., an improvement in depressive symptoms) than did IC participants (beta = 2.8, p < 0.01), no such difference was noted in the VA sample (p > 0.05). Mean costs were $D6,338 for VA IC participants; $7,007 for VA ESR participants; $3,657 for non-VA IC participants; and $3,673 for non-VA ESR participants. Although the cost-effectiveness planes suggest some uncertainty about the cost-effectiveness of the intervention, more than 75% of the bootstrap draws were considered cost-effective due to a decrease in total costs for IC in the full Veteran's Affairs sample.

The findings indicate that IC is likely to be a cost-effective intervention in contrast with ESR in the Veteran's Affairs setting. In the non-Veteran's Affairs settings, IC is not a more cost-effective intervention in comparison with ESR. In the VA setting, the greater clinical improvement associated with IC coupled with the variation in costs and outcomes were such that IC was determined to be more cost-effective than ESR with a probability of 73-80%. Among non-VA participants, the lower clinical outcomes combined with no discernable differences in costs translated with a low probability that IC was more cost-effective than ESR, at any of the estimated values of clinical improvements. This suggests the importance of clinical setting in determining the clinical and cost effectiveness of IC for mental health.

Our analyses were restricted to a six-month period, based on self-report, and did not include societal costs related to lost productivity and future costs.

These results suggest that general integration has its advantages and, when such integration exists, further integrating behavioral health care into primary care might be considered as one way to improve depression in elders. The finding that ESR may be cost effective in some settings is also policy relevant. Further research is needed to analyze the components of the costs of ESR in non-VA settings and the effectiveness of IC in VA settings.