How the Recovery Capital Index Predicts Recidivism and Housing Stability 

The Recovery Capital Index, by design, does not predict sobriety, but it can supply indications of return to use. When the data comes together with an individual’s story, we can design effective interventions based on that predictiveness. 

The number one question we get asked about the Recovery Capital Index (RCI) is, “Does it predict sobriety?”

Answer: No. But it can supply indications of a problematic return to use or other behaviors.

We understand why this is the question so many ask. It has everything to do with the traditional view of addiction; that recovery from addiction is synonymous with abstinence. Clinical care and society have long measured success as prolonged abstinence. So, it makes sense that most wonder about whether the RCI predicts sobriety. 

Designed With Intention

The truth is, we rejected this predictive premise from the outset of developing the RCI. We had anecdotal and observed evidence in our peer coaching practice of people’s lives continuing to improve, despite returns to use – either momentary or maintained.

But we also struggled with the nature of the data being used to create a predictive model.

The RCI is a self-reported, patient-driven assessment. If we were going to monitor abstinence, that would have to be self-reported too. As a non-clinical peer coaching provider, we were not interested in adding breathalyzers and urinalysis. We wanted to remove this element from the equation – which for many, was a barrier to entry. 

The more experts we talked to, the more convinced we were to focus on designing a strong and reliable measure of change. Predictiveness would come with more data and more time.

The Dangers of Prediction

We also recognized another critical fact. Introducing a predicted outcome in an uncertain and extraordinarily complex process of change is fraught with danger.

Imagine a patient completing the full Recovery Capital Index. They get a particular score. Then the provider says to the patient and their family, “Based on your RCI score, we predict sobriety at 12 months with high certainty.”

Everyone relaxes. There is some certainty for the first time in a long time. Mom and dad feel relief.

Skip ahead a few months. Our patient has been progressing nicely. Her RCI scores have been going up and her overall quality of life has been improving. But our patient has a difficult weekend and returns to use. It’s a one-time, isolated event. The predicted outcome was not met. Faith in the provider, in the patient, and in the data is compromised.

Just like that, the credibility of addiction treatment continues to be challenged.

We Cannot Skip Steps

We want to be “data-driven.”  We want to be like the rest of healthcare. We’re adding sophisticated electronic medical records, mobile apps, and wearables to a field that still operates in spreadsheets and paper forms.

We have the data science to build and apply predictive models. But just because we can do something doesn’t mean we should.

Responsible Predictiveness

The leap to predictiveness is a big one. So what,responsibly, can we do?

This very question was presented within our work in Palm Beach County. One of the agencies using the RCI was providing peer support to individuals coming out of incarceration. Participants were engaged in peer support for at least 90 days. RCI assessments were completed at initiation and every 30 days after.

The County commissioned a researcher to analyze the program. Among other questions, Dr. Heather Howard asked,: “What elements of recovery capital predict recidivism or housing stability?”

The RCI predicted housing stability at 90 days. Individuals with higher Family Support and overall Cultural Capital scores remained housed (p < .002 and <.001, respectively).

When it came to recidivism, Access to Healthcare and Values predicted rearrest among this population (p < .015).

What this kind of data allows us to do is design interventions based on results and predictiveness. For example, if a client enters a program directly from being incarcerated, the baseline assessment can provide insight into whether the person is at risk for rearrest. If Values and Access to Healthcare scores are below a particular threshold, specific instructions can be presented to the provider. Those instructions might be to connect or refer the client to a primary care physician, and to work with the client through a values exercise.

The same is applicable to housing stability. If the client has low Cultural Capital and low Family Support, the provider’s job is to help build up those areas of recovery capital.

From Data to Prediction

By using a comprehensive recovery capital measure, we can shift from binary indicators of success. Sobriety can be an individual goal or outcome, but improved quality of life requires more than abstinence. The RCI can provide insight, or predictiveness, into which areas may mitigate poor quality of life.

So far, we have uncovered some recovery intelligence for recidivism and housing stability for justice-involved individuals. As we continue our intentional approach of measuring recovery capital, we’ll gain greater predictive insight. At the same time, we will take great care with what we can predict, for whom, and what expectations that might create with patients, providers, family members, and communities. 

Recovery intelligence helps us contextualize data into a meaningful and actionable story.  Discover how Commonly Well provides impactful recovery intelligence through patient-driven analytics, behavioral insights, and outcomes for addiction treatment on our blog.