Medication Algorithms Trump Treatment-as-Usual for Mentally Ill in Texas Study

Between 1997 and 2003, the Texas Department of Mental Health and Mental Retardation joined with the state's medical schools and universities in evaluating the clinical and economic impact of an algorithm-guided treatment package on seriously mentally ill patients treated in the Texas public mental health system.

This multiphase initiative, known as the Texas Medication Algorithm Project, was the first effort by a state to create and evaluate medication algorithms for patients in a public mental health system.

In this phase of the project, researchers assessed the clinical and economic impact of algorithm-guided treatment compared with treatment-as-usual in a sample of 926 patients served in public mental health centers.

Key Findings

  • As summarized their findings in articles in the Journal of Clinical Psychiatry, Archives of General Psychiatry, Schizophrenia Bulletin, and in reports to RWJF:

    • Patients with a history of mania or bipolar disease who were treated with medication algorithms experienced a larger initial decrease in the overall severity of psychiatric symptoms compared to patients receiving treatment-as-usual.
    • There were no differences between the two groups of patients with bipolar disorder with respect to depressive symptoms.
    • All patients with major depressive disorder improved during the 12-month study period, but patients treated with the algorithm package had significantly greater reductions in symptoms and improvement in mental health functioning than patients receiving treatment-as-usual.
    • Treatment with the medication algorithm had its major effect on depressed patients within the first three months but continued to exceed the effects of treatment-as-usual for the entire one-year study period.
    • Substantial symptoms of depression for patients with major depressive disorder remained, even among patients who benefited from algorithm-guided treatment.
    • For patients with schizophrenia, treatment with the medication algorithm produced better symptom reduction than treatment-as-usual, a difference that was statistically significant but clinically modest.
    • Based upon change in symptoms and mental health care costs, cost-effectiveness varied depending on the disorder being treated.
      • For major depressive disorder, while clinical outcomes were better in algorithm treatment, the one-year treatment cost for improvement was somewhat higher.
      • For bipolar disorder, both clinical and cost outcomes were better in the algorithm group.
      • For schizophrenia cost-effectiveness did not differ between algorithm-based care and treatment-as-usual.