On a Quest to Select the Right Site First Time

• Pfizer, Inc.

The global investigative site landscape has reached an inflection point. Mergers and acquisitions, and companies sharing and pooling data, are some of the strategies aimed to make site selection more effective, yet rumor has it that sponsors and CROs are not satisfied with site identification and site selection efforts. Can we challenge the status quo to truly develop a more efficient approach to select the right site the first time? At Pfizer we are on a quest to answer this question. As part of Pfizer’s overall strategy to put innovation in the forefront of our business practice across the planning and execution of our trials, Pfizer has invested in areas of Protocol Design Optimization, Strategic Feasibility, Enrollment Modelling and Patient Recruitment and Site intelligence and Selection to become a key part of our overall Study Optimization framework. Following this organizational effort in 2018, after almost two years of a cross functional effort that assessed the internal and external landscape, we created the Site Intelligence and Selection team within our Global Product Development organization to drive this paradigm shift.

Our methodology allows Pfizer to start the site identification efforts earlier than in the traditional model, which enables a much more proactive, strategic and in-depth site intelligence data analysis. This earlier start allows us to segment our efforts to assess site footprints and identify key sites we want to nurture a relationship with at the portfolio level, new sites we have never worked with and the investigators of the future. To expand and diversify our investigator pool we put special focus on the latter. These are investigators that while they may not have proven experience in the indication, have key attributes that could potentially enable them to be a top site with the right level of support and training. This strategy is especially pivotal for our organizational patient diversity efforts, as sites located in underrepresented and or underserved populations traditionally were underrepresented in study protocols due to their lack of experience in clinical trials.

To harmonize the analysis of site intelligence and optimize the identification of the right sites the first time, we have taken into consideration multiple indicators, which we normalized within the country study and indication settings, to holistically assess the suitability of a site according to the following categories and criteria:

• Study Fit: We assess multiple key indicators of Start Up and Sites Workload.
• Site Performance: We assess multiple key indicators of Site’s Historical Performance and Quality relevant to the study indication/design. Quality being a core pillar of our site recommendations, as we are committed to ensure that our studies are delivered with high quality for the safety of our patients and the value of our science. To that end, we have made quality as a dependent variable of site performance and with the highest weight when selecting a site.
• Site Experience related to the study indication/design: Indicators related to studies conducted in the indication of interest, frequency of use in sponsored clinical trials, clinical trial site activity over time.

Additionally, our site placement assessments are plotted and geomapped on multiple dimensions, optimizing the analysis and linkage of key site attributes that allow us to enhance the prediction of a site’s ability to succeed or fail in executing a clinical trial. We look at:

• Site capacity: Number of studies they are conducting over time as reported by clinical trial public registries correlated to patient access and past performance.
• Sites Access to patients: Leveraging public census data and real-world data.
• Site Networks: History and type of collaborations amongst sites, available through the review of publications/information available in the public domain.

The aggregation of site intelligence happens first at the facility level based on a set of reference studies (similar studies completed in the past). We have learned through the development of multiple analysis that this methodology yields greater statistical significance because of the amount of data points available for the facility versus the investigator. Once we have identified the facilities targeted in the countries of interest, we review all the Investigators available at those facilities and perform another layer of assessments to recommend the most adequate investigator based on an algorithm that weights and assesses a set of key specific requirements for the study. The results of this effort yields what we believe is the most ideal pair of facilities-investigators.

The final endorsement of the facility and the investigator happens in a collaborative effort with our country stakeholders, who provide that local know-how not captured in the centralized site data repositories we review at the outset to assess site feasibility.

Site intelligence is key but is just a piece of the complex puzzle of site selection. Site outreach is another area we are attempting to optimize in this space. It is well known in the industry that when looking at the level of burden associated with clinical operations, completing site feasibility surveys ranks on top as one of the most burdensome activities reported by sites. To solve for this inefficiency our innovation efforts with regards to site engagement have always had this question in mind: How does a world without a site feasibility survey looks like? We are making good progress to solve for this question. Based on our more robust site intelligence methodology we believe we are now in a better position to identify the right sites upfront and trust our data, which reduces the traditional over reliance on site feasibility surveys and allow us to waive the site feasibility survey for those sites with the strongest evidence. In addition, with the access to more data points with regards to site intelligence we can customize the site engagement experience by site profile to only ask those questions aimed to fill in data gaps. We do this by leveraging a workflow and branched logic technology that has reduced the number of questions in our site feasibility surveys and substantially improved cycle times. We believe in a concierge type of approach when engaging with sites to assess their interest and capabilities to participate in our clinical trials.

Additionally, we have piloted with success a new approach where our site feasibility engagement is performed in conjunction with our protocol development efforts. Our site engagement in this instance is performed with two objectives: to gather insights from investigators with the objective to optimize the operational design of our clinical studies and get earlier buy in from them about their interest to conduct the proposed clinical study. The end result could have not been any better: we have observed an increased investigator motivation to participate yielding a 100% response rate from those approached. We have been able to optimize our protocol designs with their operational savviness and we have been able to reduce the time it takes to select a site allowing us to shorten start up cycle times. Scaling up this approach is one of our newest adventures.

As “intrapreneurs” we have adopted a startup approach to help us navigate on our quest to select the right site first time, with all our processes being change agile. We put special focus on aiming for excellence, breaking down silos and information clutters by promoting a 360°-learning approach through an active forum where key internal and external stakeholders actively participate.

How do we measure if we are on the right path? We have defined key primary indicators we are continuously measuring to assess if we are on the right path to shift the paradigm of site selection:

• Site Contact versus Selected Ratio: We don’t believe in an industrialized/mass mail outreach to investigators. On average we aim to approach no more than two sites for every one targeted for selection. This is enabled by our site intelligence methodology and our more proactive and strategic approach, which reduces the inefficiencies generated in the past and helps us to put the right focus on selecting the right sites.
• Site Non-Response Ratio: In the past the percentage of non-responders was a metric of concern for every study team. Presently our non-response metrics is two-fold lower than what we observed in the traditional site feasibility model. We believe that being more targeted in our site outreach efforts allows Pfizer a more personal and concierge approach, ensuring that every site engagement is followed up with the right level of attention until a resolution is achieved.
• Reducing the percentage of non-performing sites: We aim to reduce the percentage of non-performing sites by double digits in the next three years. This will be enabled by our site intelligence methodology, the building of stronger predictive algorithms, the technology advancements we are using/promoting, and increased investigator motivation based upon our more personalized interactions with sites.

What’s next? While we have been busy evolving the traditional model in site intelligence and selection, we have been working in parallel on what we believe will be the revolution in this space. We have been testing a large range of different hypothesis through our Machine Learning/Artificial Intelligence models. Partnering with business intelligence leaders and data scientists at Pfizer we are developing these models through what we call the dating game, where unsupervised learning models are assessing key site attributes significance and correlation. At this stage we have been able to build stable and viable models that can help us mitigate the risk for site nonperformance. We are also making good progress to build a larger range of predictive algorithms tools for study teams’ consumption in other areas related to protocol development and execution that have an impact on selecting the right site the first time. Keep an eye, rest assured more news is on the way as we advance in our quest!