One tension in the new thinking about program "fidelity" is between the stipulation that a developer’s specifications are followed in order to guarantee effectiveness, and the need for practitioners and local communities to feel they “own” what they are being asked to deliver.

A balance between authority and autonomy has to be struck – and autonomy and ownership usually involve a degree of license to make changes.

On both sides of the equation, these difficulties are acknowledged, but so far there have been few attempts to establish exactly what changes to an intervention communities are most likely to advocate and whether they will tend to strengthen or weaken a program’s effect.

Results from a recent study of violence and drug prevention programs in schools in northern California by researchers from Berkeley suggest over three quarters of teachers’ and students’ adaptations would have been given the green light by the programs’ developers.

The team spent two years monitoring implementation of Project Towards No Drug Abuse(TND) and Too Good for Drugs and Violence in schools. Both well regarded and and gradually advancing through the US accreditation system, the programs were being introduced as part of a new health curriculum.

Staff were given extensive training. Once the programs were running, they were interviewed and their classes were observed on numerous occasions. Participating students were invited to take part in focus groups.

All of the adaptations and suggestions for change that emerged were later assessed by the program developers. Were they likely to undermine the model and so jeopardize the effectiveness of the program?

In the main, teachers tried to integrate details from everyday life to bring the curricula to life. For example, they shared the detail of how they dealt with ordinary crises, such as a car breaking down, or they brought in newspaper articles to illustrate points made in the program. “Yeah, that’s what good teaching is. You find something that’s happened today and then put it in there,” explained one teacher.

Teachers also made a few changes to the language - to make it more intelligible and relevant to their students. More worryingly, they cut out sections they felt were less important to save teaching time, and they changed the order of the classes in response to the mood of the children.

During interviews teachers acknowledged the tension between implementing with high fidelity and providing effective teaching. They generally followed the script on the first run-through to get to know it, and they added their own touches as their self confidence increased.

Most of the curricular improvements they suggested were linked to teaching practices. For example, they were keen to introduce more group work and more interactive sessions than the program specified.

Students, on the other hand, were primarily interested in “surface” adaptations, to make the program more realistic for their age and cultural backgrounds. Their alterations were generally minor but they felt they were critical if the sessions were not to be considered boring.

The developers gave the green light to over three quarters of the adaptations and concluded that students’ suggestions were no more likely than teachers’ to threaten the logic model.

As to where the roads divided: students recommended encounters with speakers who had experiences of addiction and recovery; the developers said they would weaken the program. They were also dubious about teachers' suggestions that there should be more information about other drugs, and more group work; the developers explained these had not been empirically tested.

Where to go from here? Ozer and her collaborators look to the medical practice of evidence “farming”, a process by which practitioners’ clinical experience is integrated into evidence based on randomized controlled trials [See: "Rejecters" make the case for practice-based evidence.

No program in the world can use language or and examples that are relevant to everyone, the authors say. More work is needed to analyze and specify the parameters that will help communities to adapt programs without diluting their effect.

See: Ozer E J, Wanis M G and Bazell N (2009), “Diffusion of School-Based Prevention Programs in Two Urban Districts: Adaptations, rationales, and suggestions for change’, Prevention Science, online preview