A Quality Improvement Process for Implementing the Texas Algorithm for Schizophrenia in Ohio

Unnecessary variability in the quality of health care and the consequent unpredictability of costs have led to the development of numerous guidelines (1,2). However, despite laudable dissemination efforts, the implementation of such guidelines is the primary barrier to improving the quality of care (3,4). A recent systematic review of guideline implementation efforts outlined a number of key barriers—both environmental and physician-specific—to physicians' acceptance of guidelines (5). The six main physician-level barriers are lack of awareness, lack of familiarity, lack of agreement, lack of self-efficacy, lack of outcome expectancy, and inertia. External barriers include patient-related factors, environmental factors, and characteristics of the guidelines themselves. The quality-improvement process developed for the Ohio Medication Algorithm Project attempted to address the physician-level and external barriers to the adoption of and adherence to guidelines.

The Ohio Medication Algorithm Project is a statewide quality-improvement initiative that utilizes the algorithms developed for the Texas Medication Algorithm Project (6,7). The original impetus for considering medication algorithms in Ohio occurred in 1999, when the executive director of a community mental health board struggled with the issue of formulary costs and medication management. This concern led one of the authors (JRS) to seek out assistance from the Texas project, which laid the foundation for the planning process in Ohio and which was funded by the state and by private sources. Simultaneously, initiatives dedicated to patient safety, consumer empowerment, and consumer recovery were under way nationally. All these factors enhanced the atmosphere for the acceptability of change in practice procedures and the eventual implementation of the algorithms (8).

A locally driven process that emphasized the needs and preferences of end users was determined to be crucial for successful implementation of the algorithms (9,10). Brief initial meetings with key leaders in the state and in local settings about barriers and facilitation led to an organizing framework for the eventual implementation of the algorithms. To ensure broad-based participation, leaders from potential implementation sites, board authorities, and consumer advocacy groups selected administrative representatives to participate in a series of planning task forces. Each task force addressed themes raised at initial awareness meetings with representatives from the Texas project about barriers to implementation of the algorithms. The task forces were mandated during the course of three meetings to come up with recommendations to address potential or perceived physician-related or external barriers to implementation.

The first meeting was designed to foster in-principle agreement with the "algorithm technology and the need for it." The second meeting involved specification of objectives, and the third was dedicated to setting dates and priorities for implementation. The task forces recommended that the schizophrenia algorithm be implemented before the two other algorithms—one for bipolar disorder and one for major depression—to avoid the "information overload" associated with implementing all three algorithms simultaneously.

The project was launched in the spring of 2000 at five sites. A half-day training session, easy-to-use laminated diagrams, biweekly teleconferences during which medical directors could interact with an expert on the algorithm (ALM, PEK, or LMC), bimonthly dinner conferences, and consultation with staff of the Ohio project were used to disseminate and maintain knowledge about the schizophrenia algorithm. The initial training and reference tools were used to address the barrier of lack of awareness and familiarity. The ongoing teleconference was used to address the barriers of lack of agreement and lack of outcome expectancy. This latter barrier refers to the fact that some physicians did not believe that improvement in patient outcomes would result by following the algorithm. For example, a physician may go immediately from a monotherapy trial to a combination therapy trial of two second-generation antipsychotics instead of trying monotherapy with an alternative agent as a result of lack of confidence in the outcome expectancy—or successful treatment—associated with monotherapy after the failure of one antipsychotic. Physicians who participated in these teleconference calls were able discuss their level of agreement or disagreement with the algorithm and the perceived benefit of the algorithm in clinical practice.

Bimonthly personal interactions with an expert, with project staff, and with other physicians who were using the algorithms were used to address the barrier of lack of self-efficacy, or self-perception of the ability to perform a task. Some psychiatrists were unfamiliar with the proper discontinuation and titration strategies when switching antipsychotic regimens or second-generation antipsychotics. These meetings allowed physicians to learn, from each other, how to use these algorithms in daily practice. A Web-based decision-support system (www.bstpractice.com) included tools for patients, families, clinicians, and executives. This system was used to support familiarity and awareness of the algorithms.

Unlike the Texas algorithm project, the Ohio project did not mandate the use of rating scales at every clinical visit, the frequency of visits, case manager involvement in algorithm implementation, or a manualized consumer education program. Reducing complexity was determined to be essential for the successful implementation of the algorithms by all task forces. Support from administrative and medical leadership and from the pharmaceutical industry removed any barriers related to access to medication and budgetary constraints on physicians' time. Quality-improvement feedback as a means of supporting internal peer review and regulatory compliance was provided to address any ongoing external barriers. The objective of the study reported here was to assess physicians' attitudes toward the schizophrenia algorithm and barriers to its implementation.