Clinical Trials: A Data Driven Feasibility Approach

• Bayer AG

Abstract

Patient recruitment is perhaps one of the key challenges to the success of clinical trials. The traditional approach to assessing the feasibility of a study is to survey sites on their projected enrollment capability. This information gathered by reaching out to sites is useful but by itself it isn’t sufficient as the sole source of information. In this paper we propose an approach for a data driven feasibility including key considerations which should be taken into account to conduct successful feasibility assessments and minimize risk.

Introduction

Patient recruitment is perhaps the biggest challenge to the success of clinical trials. Inability to recruit subjects in target timelines results in an extensive cost and resource burden on study teams. This reduced efficiency due to the extended recruitment time is a common problem for all drug sponsors costing them millions each year. Approximately 80% of clinical trials fail to meet enrollment timelines and one-third (30%) of Phase III study terminations are due to enrollment difficulties.¹

Key reasons for recruitment delays are –

• Inability to select sites that will deliver the patients.
• Uncertainties in estimation of number of patients from sites- “The number of patients predicted by investigators typically falls by up to 90% at the start of a study.”²
• Improper estimation of time required for patient enrollment.
• The protocol design.
• The competitive landscape.

Sponsors wish for greater trial predictability where enrollment forecasts are realistic and achievable. Given that there is no optimal solution to this problem, in this paper we present an approach which we believe allows for informed decision making through a data centric approach which is adaptive to the needs of each study. Investing in a robust feasibility process based on early planning enables sponsors and other involved parties to obtain a realistic assessment of the capability of executing a planned clinical trial.

The Challenge/Current Approach

The traditional approach to assessing the feasibility of a study is to survey sites on their projected enrollment capability. Investigators are provided details of the inclusion/exclusion criteria and protocol summary and are then asked to complete questionnaires to estimate the number of patients they believe they can recruit for the trial based on the given selection criteria. This information gathered by reaching out to sites is useful but by itself it isn’t sufficient as the sole source of information:

• Most investigators don’t check the inclusion/exclusion criteria of the study against a patient database. They only provide a rough estimate.
• Investigators may overestimate the number of patients they expect to recruit in the hopes of being selected to conduct the study. In other cases, investigators may underestimate so that the target enrollment for the site is easily achievable.

Sponsors analyze the gathered patient count estimates and use their own experience to suggest the number of patients they might expect per site per month. If the sponsor has limited experience in the therapeutic area/study type, they are completely dependent on the gathered estimates. In many cases, this leads to over-optimism and unrealistic targets. Many times, sites get selected due to having worked with the sponsor in studies in the past. This neglects that each study is different and thus have differing requirements. Site selection must be based solely on evidence that a site is expected to perform well.

Unfortunately, this planning isn’t given sufficient emphasis, and only gets emphasis when an enrollment delay is in full swing. Detailed early planning can be time consuming and costly but the cost is justified as it can prevent such a crisis from even happening in the first place.

Proposed Approach/Process

Trial feasibility is an art supported by robust data and informatics to increase the degree of confidence. The feasibility process should be designed to be adaptive. It should start early and should be customized to the specific needs of therapeutic areas and studies. The key attributes of a robust feasibility approach include –

• Clearly defined triggers:
• Feasibility should be started early and has clearly defined triggers. This trigger should not be linked to availability of specific documents but rather to the availability of the minimum necessary information to begin feasibility. This allows for earlier planning and an earlier start to feasibility activities.
• Adaptive approach:
• Feasibility should be managed using an evolving Feasibility Plan that begins during Clinical Development Plan (CDP) progress and will outline the specific steps required for the program and its studies.
• This plan defines the depth and breadth of feasibility along with planned timelines for feasibility. The aim is to plan early and allow the approach to be tailored to each program/study.
• Data-driven:
• Decisions should be driven by all available data where selection of sites is based on past performance (historical, internal and procured data) and suitability (assessment via surveys).
• Collaboration with countries:
• Countries should be involved from the beginning to enable effective resource planning. Face-to-face interactions with investigators are recommended to help drive a partnership throughout the process.

A three stage approach is proposed to conduct trial feasibility.

 Figure 1. Alignment of feasibility activities with study milestones

Early Indication Feasibility

Feasibility should be initiated at a defined trigger point and should begin with a planning stage. During early indication feasibility the main objective is to develop the feasibility plan for the program/study. This includes the extent of feasibility required based on prior experience in the indication and agreement on feasibility timelines.

Intelligence should be gathered through all available sources (internal and external) to draft an initial global investigator list. This serves as a long list to be further refined through the feasibility process. It is also important to involve countries here early in helping develop the initial investigator list as well as developing an outreach plan to engage with the investigators.

Protocol Feasibility

Protocol feasibility should begin once a draft protocol is available to conduct feasibility. The objective of this next phase is to obtain feedback from selected countries and inves-tigators on the protocol and assess the protocol’s feasibility. Also, the need to conduct Evidence-based-Feasibility (EbF) should be assessed in this phase.

The questionnaire for protocol feasibility needs to be designed well to allow open feedback and yet provide comparable feed-back from the sites. The sponsor needs to find the right mix of structured and unstructured questions in their surveys to keep the survey short and yet enabling sites to provide the right information.

The results from the survey should be collated and reviewed for feedback on protocol design and other feasibility challenges. Scenarios need to be modelled to plan the number of countries and sites needed to conduct the trial.

Site Feasibility

At the conclusion of protocol feasibility, we move to site feasibility where the objective is to reach out to investigators and select sites for the trial.

Results from the site feasibility enable development of the study enrollment plan and agreement on the final site list for the study.

Recommendations

• Start the feasibility planning process early. Feasibility planning can be started once the stable CDP with key information is available. Availability of stable primary endpoints, inclusion/exclusion criteria, study design, number of patients, duration of treatment and information on comparators be considered as a good trigger. Defining the depth and breadth of feasibility approach early would allow the team to plan better and design an approach solely for the program and portfolio of studies. During the planning stage, a gap analysis can be conducted to understand the availability of internal knowledge/experiences, extent of external data required, prevalence of the disease and also the competitive landscape. The feasibility plan should be flexible but structured with clearly defined steps and timeline. The plan should evolve as the protocols get finalized. During the planning stage the cost of involving external suppliers should also be considered. Personnel with clinical and operational expertise should be involved in the planning.
• Once the approach is defined it might be useful to assess if the program outlined in CDP is feasible with the proposed timelines and procedures. This assessment can be conducted with the available internal historical data coupled with benchmark trials/competitive intelligence and mathematical modeling.
• Mathematical modeling and simulation can be used during the early stage for better operational planning. Modelling enables the study team to explore multiple options and potential impacts on the study timeline. For example, how many countries and sites are required to meet the study timelines? What would be the potential impacts on timelines if more countries/sites are added? If we add additional sites would it reduce the study timeline significantly? What would be the impact on timelines if we invest on a recruitment campaign? With the mathematical modeling, multiple scenarios can be built to explore all options. Past performance of countries and sites, cost, start-up timelines can also be factored in for country and site selection. Often modelling can help to assess multiple options – for example cost of opening up new sites versus extending the timeline by a few months. Conducting modeling early can help to develop a realistic plan, it should also be repeated once the study specific feasibility data (e.g. predicted rate of recruitment) are collected from the sites. Modeling can also help in predicting recruitment performances of sites during the study provided plan versus actual data is available.
• Based on the CDP level information an initial list of potential sites can also be generated for further assessment. Past enrollment performances, access to potential patients, infrastructure, and quality of data generated in the past and availability of trained staff can be used to develop a potential site list. However, depending on the needs of the study selection criteria can be adjusted through assigning appropriate weight to the most important criteria. For example, access to potential patients is more important than taking part in the clinical trials in the past for a rare disease or ultra-rare disease.
• Develop a robust questionnaire for protocol feasibility focusing on medical and operational aspects of the study. Try to limit the number of questions to 30 if possible. Questionnaire should be easy to navigate and structured so that users can easily enter the data. Any questionnaire taking more than 30 minutes to complete for a site may reduce the overall response rate. The language should be carefully drafted. The survey populations should be selected carefully and should be representative of all the potential countries.
• The use of Evidence-based-Feasibility (EbF) explores the opportunity to assess inclusion/exclusion criteria on anonymized real-time electronic medical records. It can be conducted remotely through selected hospital networks using a dedicated platform, third part vendors or on-site from a sample of the target patient population. The objectives of the evidence based feasibility include identification of the proportion of the population that matches the protocol, potential impact on the availability of the patient pool if some of the inclusion/exclusion criteria are changed, exploring country specific clinical practice/standard of care, patient pathways, referral networks and potential recruitment rates and challenges. The process is usually done prior to protocol finalization, so changes can be made proactively to the protocol, minimizing amendments. EbF can also be used to support “root cause” analysis of active studies experiencing recruitment issues and also for improved site selection. The results of EbF should be used carefully with caution as the suggested numbers are derived from a small sample population. Simple extrapolation may result in an accurate prediction.
• Maximize number of site visits during the site feasibility to get better quality data and face-to-face visits would allow the sites to clarify information quickly. Allow adequate times for countries to conduct site feasibility. Involve a country medical expert early in the feasibility process to get the local knowledge in the questionnaire development and selection of the right sites.
• Develop recruitment/retention strategy for the study, country and sites as early as possible which is funded and staffed either at study or country level depending on what strategy is implemented. Develop country/site specific outreach and engagement materials. Ensure you have Master Service Agreements and rate cards agreed with potential recruitment vendors for a quicker deployment.
• Track enrollment closely on a regular basis and make predictions so that adequate strategy can be deployed before the enrollment target is missed. The longer the screening phase, the closer recruitment needs to be monitored (run-in monitoring). Define enrollment predictors upfront and test your assumptions for your screening failure rates.
• Consider early closure of sites which don’t recruit patients.
• Consider closing the randomization at a certain point to avoid over-recruitment and communicate this appropriately to sites.

Conclusion

Data driven feasibility assessment can improve the probability of getting the right sites to improve the recruitment performance. It allows for informed decision making where all available information is taken into consideration. We must remember though that studies can fail to recruit due to various reasons even though predictive analytics is used, but quality data and a data driven feasibility can help minimise risks.

References

1. Clinical trial delays: America’s patient recruitment dilemma (2012)
2. Dr Louis Lasagna: Lasagna's law http://www.pmean.com/11/lasagna.html