Best Practices: Disseminating Best Practices for Bipolar Disorder Treatment in a Correctional Population

Two inmates were transferred to other prisons and had to be withdrawn from the project because these other facilities were not approved for protocol implementation. Four inmates dropped out of the study because they were released by court order. All of these inmates were lost to follow-up. Another administrative issue hampering algorithm implementation involved inconsistent blood drawing procedures. This inconsistency was partially due to lack of resources. We had to frequently educate the staff about the importance of timely blood draws from both an efficacy and a safety standpoint.

CDOC formulary restrictions also complicated algorithm implementation. Agents such as olanzapine and aripiprazole were not on the CDOC formulary. Per CDOC policy, psychiatrists were able to access nonformulary medications only after failed trials of formulary medications. Because of this policy, the nonformulary medications were considered at later stages during the implementation, at times inconsistent with the algorithm recommendations. It is important to note that clinicians seldom used the nonformulary medications because they were able to achieve adequate response using formulary medications.

These issues were related to assumptions and expectations of inmates or clinical support staff or both.

Issues related to inmates and clinicians. Inmates, at times, refused to consider certain medications (for example, lithium) without any appropriate reason and at times because of hearsay about the medication. This phenomenon appears to have occurred much more frequently in this setting than in our noncorrectional outpatient settings, possibly because of constant proximity to other inmates, a lot of free time, and gossip about medications. We had to educate inmates about the importance of considering all potential options recommended by the algorithm and describe it in the context of risk-benefit ratio of all available options. Furthermore, inmates frequently requested switching or adding medications in the absence of any side effects and often before completion of an adequate medication trial (dosage or duration) per algorithm recommendation. Occasionally such requests were tacitly endorsed by the correctional support staff. Study clinicians faced difficulties adhering to algorithm-recommended critical decision points ( 10 ) because of these requests. Inmates and occasionally correctional staff needed education and reassurance about the potential benefit of the ongoing, algorithm-recommended medication trial. We also had to educate inmates in individual sessions about the importance of an adequate trial of a medication recommended by the algorithm. Study clinicians were, in general, successful in reassuring inmates and staff, and for the most part, inmates experienced improvement if they continued on the trials.

Issues specific to clinical and support staff. At baseline, we often encountered simultaneous use of three or more psychotropic medications from different stages of the algorithm. Frequently these medications were used at dosages that were less than the recommended ones. This baseline polypharmacy was partially due to the traditional symptom-driven management of bipolar disorder. To address this issue, we had to educate clinical support staff about the importance of providing evidence-based care, of considering the efficacy-safety ratio, and of focusing on the treatment of illness rather than treatment of individual symptoms. We also educated them about how evidence-based algorithms can be used to accomplish these objectives. Another issue complicating algorithm implementation was the fact that utilization of specific classes of medications at baseline was often inconsistent with current evidence for the treatment of bipolar disorder. In several cases, medications recommended at the first algorithm stage had not been tried. All of these factors complicated algorithm implementation because study clinicians frequently had to move backwards in the algorithm to start from the first stage. However, it is important to note the improvement trend that occurred during the implementation of the TIMA algorithm. Additionally, there was a considerable decrease in utilization of antipsychotics and antidepressants and an increase in anticonvulsant medication utilization, changes consistent with algorithm recommendations.