How to be data-driven in support of trial diversity goals

Sponsors and CROs want trials to be faster and cheaper. They also want them to be more diverse. This requires data. Unfortunately, the industry hasn’t historically tracked and reported on the diversity stats we need to make more informed decisions.

We cannot manage what we do not measure.

We believe that change begins with transparency across recruitment channels and sites. So here are the top two areas where we need to be more data-driven while planning diverse recruitment programs.

1. Evaluate and prioritize recruitment channels based on the representativeness of their patient population.

Traditional recruitment channels are not representative. It is difficult to tell just how skewed these recruitment channels are, however, because there is almost no data on the utility of specific channels to recruit diverse participants. Without transparency into representativeness on a per-channel basis, sites, sponsors, and CROs are left guessing what will work to enhance clinical trial diversity.

While there is minimal available data on by-channel representation, current recruitment statistics indicate that the status quo is not effective at recruiting diverse participants. Today, it is too hard to find trial recruitment information because websites like clinicaltrials.gov and other common tools are not accessible for most patients to navigate. When patients do find trials, the information is often overly complex and almost exclusively presented in English. Therefore, if patients aren’t already seeing a provider who knows about research in the area, they are much less likely to find out about research opportunities. But these more accessible channels like physician referral and in-clinic recruitment target white, affluent participants because they are more likely to see physicians within major research networks, in higher socioeconomic status areas.

To solve this problem, sites, CROs, and sponsors should evaluate and prioritize patient population representativeness during recruitment vendor selection. Stakeholders must prioritize channels that are best at driving diverse representation, but it is currently difficult to rank channels based on representativeness because there is no established practice for data collection or reporting on the ability of various channels to recruit diverse participants. There is also minimal empirical data on the utility of alternate formats to recruit diverse populations. Sponsors and CROs suffer as a result because it is risky to invest in new recruitment platforms without assurance that they will target representative populations. Since there is no data on the representativeness of existing platforms, it is impossible to compare new solutions.

The first step forward is to evaluate and prioritize channels based on the representativeness of their patient population. Sponsors, CROs, and individual sites should collect data on the number of enrollments of a demographic group sourced from each channel. For example, data on demographics sourced from physician referrals vs. social media outreach vs. public access websites for a specific disease area in a region can start to inform a larger ‘map’ of the channels that recruit diverse participants. Over time, larger sponsors and CROs can become leaders in the recruitment space by developing a disease-specific database on recruitment platforms. These databases should be interactive and responsive to change over time.

If sponsors begin requesting and collecting this data from recruitment vendors, these vendors will quickly make this information more accessible. More accessible representation information will very quickly facilitate cross-vendor comparison, which will in turn incentivize vendors to optimize their own tools for diversity. Many stakeholders may find that they actually have already been collecting this data in different forms. A small investment now can pay dividends down the road in more efficient, diverse recruitment.

Stakeholders should also build redundancy into their representative recruitment strategy, so as not to be overly reliant on one channel to ensure participant diversity. Recruitment teams should start by ensuring that resources are default multilingual, contain representative imagery, and adhere to the Web Content Accessibility Guidelines. They must also leverage partners who have demonstrated that their patient population is nearly representative of the census, such as patient education platforms like Power, disease-specific advocacy groups, and diverse patient support groups.

2. Create site-level demographic reporting systems.

There is currently a lack of metrics and systems for reporting site-level demographics. Most sites do not report on recruitment demographics as a key performance indicator, nor do sites and sponsors expect this granularity of detail.

The deficit in site-level demographic reporting creates a self-reinforcing cycle. The lack of standardized metrics for reporting patient demographics means that sponsors cannot make effective comparisons between sites. The lack of ability to make comparisons means that individual sites don’t have the incentive to publish their demographic data. The lack of available data means there’s no impetus to develop standardized metrics for reporting said data, so sponsors continue to be unable to make effective comparisons between sites.

The deficit in site-level demographic reporting creates a self-reinforcing cycle. The lack of standardized metrics for reporting patient demographics means that sponsors cannot make effective comparisons between sites. The lack of ability to make comparisons means that individual sites don’t have the incentive to publish their demographic data. The lack of available data means there’s no impetus to develop standardized metrics for reporting said data, so sponsors continue to be unable to make effective comparisons between sites.

Sponsors who wish to drive change in their field need to break this cycle by creating defined processes to report the demographics of patients at every step in the enrollment funnel: interest, prescreening, screening, enrollment, randomization, and trial completion. Sponsors should also request data on the demographics of patients who visit the site regularly if the site also operates as a clinic or pharmacy. This data, bundled into leveled reports, will help sponsors develop a clearer idea of where racialized participants fall off in the funnel. The deficit in site-level demographic reporting creates a self-reinforcing cycle. The lack of standardized metrics for reporting patient demographics means that sponsors cannot make effective comparisons between sites. The lack of ability to make comparisons means that individual sites don’t have the incentive to publish their demographic data. The lack of available data means there’s no impetus to develop standardized metrics for reporting said data, so sponsors continue to be unable to make effective comparisons between sites.

Ultimately, more diversity and inclusion data analytics is always useful if there is a minor investment in analytics. If a few sponsors lead the way in requesting this data from their sites, they can quickly incentivize a culture shift where this kind of data collection is common across all sites in the country. Once sites are attuned to the shift, they may also choose to analyze their own site-level data goals and modify practices to ensure they are recruiting the most representative patient populations.

Today, sites are frequently hesitant to share information that may negatively impact their chance of getting a study. If their recruitment data is not representative, they may push back at the idea of mandated sharing to win trials. Sponsors should frame this diversity data collection as a method of helping sites. While this data may be a good tool to evaluate site-level success, it is also important as an analytical tool to see where the industry as a whole can do better.

Sites should use this culture shift as a push to leverage diversity analytics when trying to win trials. As this practice becomes more common at the site level, it will naturally incentivize other sites with similar profiles to share diversity performance indicators, and make it more feasible for sponsors to require these metrics when building site networks.

Taking the first steps

Here are a few steps sponsors can begin with to kickstart the process:

1. Choose metrics: Sponsors should begin by outlining the recruitment channel and diversity metrics dashboards that are easiest to collect and most closely align with their recruitment goals.

2. Establish data collection processes: The next step is to create processes to easily analyze and update this data across time and geography.

3. Budget for diversity: Finally, sponsors should create sample diversity and inclusion budgets that outline specific time and financial goals for diversity data collection over the next five years.

This research is part of a larger report on achieving diversity in clinical research. Read the full report here.