A regal location in central London was the venue for a meeting of minds to distill the learning from both sides of the Atlantic about taking evidence-based interventions to scale. David Olds was there to share his experience of taking the Nurse Family Partnership (NFP) - the home visitation program helping first time vulnerable mothers and their children through pregnancy and infancy - to scale. He was joined by Kate Billingham and Ann Rowe, who are taking forth the mantle of wide-scale implementation of the program in the UK (known locally as Family Nurse Partnership (FNP), as well as Jacqueline Barnes who is leading the evaluation of implementation. A select group of senior policy-makers, investors, academics and program developers from the UK joined the discussion.

According to Olds, “the first step to wide-scale implementation of any evidence-based program is to first make sure that the evidence-base is solid”. Replication - the process of taking a core idea that is proven to work, often in tightly controlled, well-funded and auspicious circumstances, then replicating the delivery of that program in ‘real-world’ settings with sustained effects - is critical before wide-spread expansion. This sentiment is echoed by Billingham when she says that “contrary to the pressing demands of such policy, you can’t run before you can walk; taking the time to replicate and see if interventions are effective across contexts is a critical foundation for expansion”. Soberingly, it took Olds over 20 years and three randomized controlled trials to achieve this with NFP.

Only when the evidential foundations of an intervention are solid should wide-scale implementation occur. To Ann Rowe, not only is such a foundation essential for policy-makers seeking confidence that investments in intervention efforts will pay off, but also, it is also critical for practitioners delivering the intervention to children and families: “the evidence-base helps instill practitioners with the aspiration and confidence to deliver the intervention as intended”.

The last ten years has seen an unprecedented growth in the number of evidence-based programs with such an evidentiary foundation, but why have so few been successfully taken to scale? The collective experience in the room, from both sides of the Atlantic, pulls out four key ingredients essential for taking evidence-based programs to the masses.

First is buy-in from policy-makers, commissioners, communities, practitioners and those children or families receiving the intervention. And not just buy-in from one or two of these stake-holders, but all of them. According to Olds, this requires leadership, organizational capacity and a firm commitment - ideally contractual - to deliver the intervention as intended. Kate Billingham and Ann Rowe argue that a license to deliver an evidence-based program is critical; it holds the deliverers of the intervention to account and demands fidelity.

Second, clear manuals and practitioner guidelines are a necessary but not sufficient compenent; as articulated by Rowe, “there is a lot more to practice than just guidelines”. As such, manuals and guidelines must also be coupled with excellent ‘technical assistance’, that is, the availability of experts on the program implementation to help set the course, fix problems as they occur and build capacity for dealing with those problems if they occur in the future. In many programs, such as NFP, high quality and regular supervision from experienced deliverers is critical.

Third are clearly specified targets around program implementation, especially regarding (i) recruitment and engagement of those targeted by the intervention; (ii) retention of those individuals throughout the course of intervention; and (iii) a sufficient ‘dose’ of the intervention. Olds says that it helps to ‘stretch’ these targets too, so that required recruitment, retention and dose rates are above the minimum levels necessary for demonstrable impact on child outcomes, thus leaving some room for error.

Fourth, and related, is the requirement to continually collect data regarding implementation quality and fidelity to the program. This not only helps ensure that the program is being delivered as intended, but also fosters reflective practice of those delivering the intervention and allows researchers the opportunity to learn more about the effective components of intervention. According to Olds, “it is this continual reflection and learning that helps ensure that interventions stand the best possible chance of improving the lives of those they serve”.

But, of course, it is not always so straight-forward in practice; as Billingham says, “it's a hostile world out there, ready to corrupt the purity of evidence-based programs in an effort to make them cheaper, more streamlined and faster to implement”.

How do we ward against this tendency to be, as Billingham puts it, so “promiscuous” when it comes to implementing evidence-based programs? Olds argues that evidence-based programs should only be taken to scale in their pure form; “anything less, and you jeopardize the potential to improve child and family outcomes, which, is not only a moral issue, but also an economic one”; why spend precious resource rolling out a watered-down intervention for which, at best, the effects are likely to be greatly diminished and uneconomical, and at worst, harmful?

That said, it is obvious listening to Olds and Billingham that a perpetual curiosity and humility is central to bringing evidence-based programs to life. Only by asking the questions about how best to improve outcomes for children, and testing emerging ideas, do we end up with programs like NFP that demonstrate such marked improvement in child outcomes.

So when asked if a more ‘streamlined, faster or cheaper’ version of NFP, might work, after 30 years of committed work, Olds still has the humility to say “I don’t know, we’d need to test it to find out”. But he is keen to stress that when it comes to taking interventions to scale, “not knowing is not enough”. There is always room for innovation and humility, but this must precede wide-scale implementation and should not occur in tandem.