Patients Recruitment Forecast In Clinical Trials

Cognizant 20-20 InsightsPatients Recruitment Forecast inClinical TrialsAccurate patient recruitment forecasts are critical to the clinical trialplanning process. Here, our recommendations on ways to remain ontarget and avoid huge losses in terms of time, effort and investment.Executive SummaryClinical trials are typically the most crucial partof a drug development cycle. After the manufacturer – a pharmaceuticals or medical devicecompany – has spent a significant amount oftime, effort and funds, it must test to make surethe drug is marketable and meets its therapeuticpromise. Any failure at this stage can set the drugmanufacturer back years.Clinical trials are a long and tiresome process.Any misses here can delay time, add costs andresult in missed opportunity. The leading culpritfor missed clinical trial deadlines is the patientrecruitment process. Patient enrollment is themost time-consuming aspect of the clinical trialprocess, estimated to take up to 30% of theclinical timeline. At the sensitive and crucial stageof development represented by clinical trials,optimizing patient enrollments with improvedrecruitment rates offers a clear advantage thatresults in time savings and reduced time to launch.Clinical trials are the testing benchmark thatmake or break a drug. But they are imperfect,due to strict regulatory protections put in placeto safeguard human subjects. The more patientsneeded for clinical trials, the greater the numberof regulatory issues that typically arise related tocognizant 20-20 insights august 2015patient safety and procedural validation. This, inturn, extends trial duration, and on a larger scaleimpacts multiple sites simultaneously. The impactoften leads to increased demand for patientrecruitment.This white paper explores different approachesfor forecasting patient enrollment, including basicguidelines for uncontrollable factors that may bequantified for better control over clinical trials.From the Beginning: Patient RecruitmentPatient recruitment services annually contributeover 5.9 billion in expenses to the pharmaceuticals industry.1 With such a significant amount ofmoney being invested, the scope becomes so specialized that there are patient recruitment serviceproviders that facilitate a variety of services toincrease enrollment in clinical trials.Like other industries that have evolved over time,pharmaceuticals and medical devices companiestypically stick to their historical approach, at leastin terms of infrastructural investments for clinicaltrials. With the correct protocol and investigatorsites, patient enrollment should be facilitated.When a a state-of-the-art site is installed, thereis no logical reason to believe that it should

be difficult to recruit patients and providetreatment there. Unfortunately, this thinking,while delivering enormous success historically,has outlived its usefulness. Given how inherentlycomplex the process is for industry, manufacturers and patients, nothing can be taken for grantedand everything must be meticulously plannedwherever possible.The industry transformation that has led to sucha complicated state can be understood in light ofthe following:2 Roughly80% of clinical trials fail to meetenrollment timelines. Approximatelyone-third (30%) of phase IIIstudy terminations are due to enrollment difficulties.As these statistics make abundantly clear, betterpatient recruitment processes can drasticallyimprove clinical trial success.What Makes Patient Enrollment aHerculean TaskPharma and medical device companies startingclinical trials are aware of the complexitiesinvolved in patient enrollment, but the toughertask is to identify the causes of obstacles. Witheach trial, a new set of conditions is available,with each condition requiring a different testingapproach. In other words, the qualitative factorssuch as location, type of disease, drug andcompany reputation play an important rolein shaping the trial. Therefore, many factorscontribute toward making patient enrollmentgenuinely difficult and arguably the most challenging step in running a clinical trial.Improper Estimation of Time Required forPatient EnrollmentAny big project with long-lasting and vital impactneeds proper planning. Plus, companies must havea reliable estimate of what can be expected. If thecompany fails here, the reality will soon divergefrom the plan. The primary problem faced duringpatient enrollments is improper expectations dueto faulty trial forecasts. In short, a trial in actualityoften takes a longer time than estimated to enrollthe required number of patients. For example,a trial forecasted to have the right number ofpatients enrolled in, say, 20 months can actuallytake 22 months to accomplish this. This mismatchcan easily lead to huge losses in terms of time,effort and investment. Any deviation from theforecast is not a good sign, and it is advisablecognizant 20-20 insightsto be absolutely certain about the trial planningprocess.Patient Behavior and Drop RateSuccessful enrollment alone does not ensure thesuccess of the entire trial. It is only the beginningstage, and there are many more uncertainties toaccount for. For example, a patient recruited isnot bound to stay with the trial. The reluctanceof a recruited patient can be attributed to a widevariety of reasons ranging from simple logistics,to fear, to any inconvenience in adhering to thesteps in the protocol, or to the invasiveness oftreatment and diagnostic procedures. This revealsyet another problematic issue: patient drop ratesfrom trials.The National Cancer Institute (NCI) has devotedefforts to research such factors, and has reportedthe following primary reasons offered by patientsfor a positive attitude and for trial participation: Recommendation or influence of a doctor. Hope for therapeutic benefit. Altruism or to advance science. Lack of other medical options. Access to leading specialists. Ability to receive cutting-edge care andthelatest treatment discoveries.Uncertainties in Estimations from SiteInvestigatorsWith all these uncertainties, sponsors often turnto the direct recommendation of site investigatorsfor estimates of patient enrollments and recruitment success. Investigators provide recommendations using projected enrollment capacities. Themost common way to collect these projections isthrough questionnaires. The questionnaires arerather simplistic, using questions focused on thenumber of patients the investigators treat who fitthe study criteria and how many they believe theywould be able to recruit for an upcoming trial.Physicians are busy professionals who are intenton working in their patients’ best interests.However, the aforementioned activities todetermine patient counts are, at best, estimatesgleaned through a summation of educatedguesses. Moreover, a physician is not bound toanswer this questionnaire, and feasibility surveysmay go unanswered at some investigator sites.The major limitation that emerges from theseestimates is that the site investigators tend toovercommit and are overly optimistic about their2

ability to supply patients. In a typical trial, morethan one-third of sites under-enroll patients androughly one-tenth fail to enroll even a singlepatient.3 Thus, site potential is unutilized andmismanaged.results in some cases; even deterministic methodsmay perform better than stochastic ones, so themethodology selection should be based on theperformance of both methods for the situation athand.Reestimation by SponsorsNon-Stochastic ApproachThe companies are aware of the “overestimation”issue and bring in their intuition and best judgmentto manipulate the survey results to reestimate thetime it would take to enroll all patients requiredfor the trial. Even here, companies may err inputting intuition over estimates, with the aim tocorrect estimates that go astray from recruitment success rates.All of these steps are part of the process, andwith experience, calculated estimates and a bitof luck can generate a fairly workable enrollmentschedule. The question is one of accuracy. Fortunately, statistical techniques are available thatcan process the rich and varied data throughmultiple scenarios to provide accurate information with confidence for various trial forecasts.The statistical methods are designed to work overthe information gathered from physician surveysand can withstand the element of interferenceengendered by unintended company tampering.The output is a baseline forecast with a highlyaccurate predicted probability of enrollmentsuccess for the intended timeframe. Given that theproper data sources and analytical steps are usedwith correct statistical methodologies, companiescan obtain the baseline estimates fairly quickly,with research steps that are performed concurrently.Today (to the best of our knowledge when thiswhite paper was written), the most populartechniques used by companies for recruitmentand supply modeling use averages and other adhoc techniques. These deterministic techniquesare not devised to account for uncertaintiesarising from: Uncertain input information. Stochastic fluctuations over time. Change in recruitment patternsIn this approach, the accuracy of the outcomesis improved through known relationships amongfactors under consideration, removing the possibility of any random variation. With this approach,a given input always produces the same output.Patient recruitment and recruitment duration aredependent on total number of patients (P); totalnumber of sites (S); number of sites enrolled atthe beginning (S0); number of patients enrolled atthe start of the trial (P0); patient enrollment rate(λ); and average site initiation rate (Ss).Forecasting clinical trial enrollment requiresestimates of patient recruitment rate (λ); and sitesstart-up timing (Ss). Patient recruitment durationand patient recruitment can be forecasted at twostages of the clinical trial: At the beginning of the trial. At the interim stage of the trial.The clinical operation department can provideminimum, maximum and average values of λand Ss and these values can be used to estimateexpected values and standard deviations of λ andSs through the Program Evaluation and ResourceTechnique (PERT).This approach assumes that sites start with anaverage initiation rate until all start enrolling.Initially, total recruitment is conditional upon thetotal sites available to enroll; it becomes unconditional and linear once all sites are enrolled.Patient RecruitmentOn or before enrollment completion of all sites,the number of patients enrolled (Pq) Aλ A*λ P0 Where A (1/2)rd2 and A* S0dAfter enrollment, completion of all sites numberof patients enrolled (Pl) Sλd Pqand ratesacross sites.In our experience, over half the companies thatare using statistical methods also fail to recruitin time. Companies using stochastic methodologies implement Monte Carlo simulation to bring inthe uncertainties factor to predict patient recruitment. Stochastic and non-stochastic approacheswith Monte Carlo simulation may provide similarcognizant 20-20 insightsRecruitment duration can be calculated as follows: If the number of patients recruited is equal tototal number of patients (P), then average siteinitiation time (Ss) would be equivalent to D. If the number of patients recruited is greaterthan P, then D would be the positive rootsolution of equation 1. If the number of patients are less than P, then3