Modeling Clinical Trial Attrition Using Machine Intelligence: A driver analytics case study using 1,325 trials representing one million patients

The amount of time and resources invested in bringing novel therapeutics to market has increased year over year with fewer successful treatments reaching patients. In the lifecycle of drug development, the clinical phase is a major contributor to this decreasing efficiency in the development of clinical trials. One major barrier to the successful execution of a randomized control trial (RCT) is the attrition of patients who no longer participate in a trial either following enrollment or randomization. To address this problem, we have assembled a unique dataset by integrating multiple public databases including ClinicalTrials.gov and Aggregate Analysis of ClincalTrials.gov (AACT) to assemble a trial sponsor-independent dataset. This data spans 20 years of clinical trials and over 1 million patients (3,175 cohorts consisting of 1,020,085 patients and 79 curated features) in the respiratory domain and enabled a data-driven approach to identify top features influencing patient attrition in a trial. Top Features included Duration of Trial, Duration of Treatment, Indication, and Number of Adverse Events. We evaluated multiple machine learning models and found the best performance on the Test Set with Random Forest (Test subset: n=637 cohorts; RMSE 6.64). We envisage that our work will enable clinical trial sponsors to optimize trial run time by better anticipating and correcting for potential patient attrition using patient-centric strategies to improve patient engagement, thus enabling new therapies to be delivered to patients more quickly.

1203. A Descriptive Analysis of an Opioid Use Disorder Care Continuum in an Infectious Diseases Clinic

Background On December 17, 2020, U.S. CDC released an advisory reporting the highest drug overdose rate on record. Kentucky ranks in the top 5 states for opioid overdose deaths. Retention in opioid use disorder (OUD) treatment is associated with decreased overdose deaths. University of Kentucky HealthCare’s infectious disease division (UKID) implemented a multi-disciplinary approach to expand access to medication for opioid use disorder (MOUD) for patients with injection drug use-associated infections (IDU-AI). This program is modelled after the Ryan White Cares Act to engage and retain patients. Methods . This ongoing project began enrollment in June 2019. Any patient ≥18 years old with IDU-AI and OUD is eligible for enrollment unless pregnant or incarcerated. Patients are eligible for transportation assistance, mental health services, and medical case management. They may start MOUD with UKID or be referred elsewhere. In this analysis, we describe our opioid use disorder care continuum and identify reasons for patient attrition and areas to improve Results Our continuum components are referral, eligible, enrolled, start MOUD, and retention at month 1, 3, and 6. To date, 533 patients have been referred. Of these, 383 (71.9%) were eligible and 150 (39%) enrolled. Reasons patients did not enroll: discharged stable (41.5%), left AMA (16.9%), declined (10.8%), deceased (6.7%), discharged to other hospital (3.6%), missed clinic visit (9.7%), hospice (1%), other (10.8%). Reasons patients declined: no reason (28.6%), refused to discuss (19.1%), no interest (14.3%), travel (4.8%), declined ID follow-up (4.8%), time limits (9.5%). Ninety-three patients have been enrolled ≥6 months; 83 are on MOUD. Sixty-seven, 29, and 20 patients were retained at month 1, 3, and 6, respectively. Conclusion UKID engages patients in OUD treatment, but retention rates are comparable to those described in non-ID settings. Most attrition occurs between eligibility and month 3, suggesting patients are most vulnerable when they consider change and start MOUD. These time points should be priority for patient engagement by clinic staff. Also our staff size struggles to meet the demand. The number of referrals is prohibitive for our small team to approach everyone in a timely manner. More programs like this one are needed. Disclosures All Authors: No reported disclosures

Patient attrition in Molecular Tumour Boards: A Review

Molecular Tumour Boards (MTBs) were created with the purpose of supporting clinical decision making within precision medicine. Though these meetings are in use globally reporting often focuses on the small percentages of patients that receive treatment via this process and are less likely to report on, and assess, patients who do not receive treatment. A literature review was performed to understand patient attrition within MTBs and barriers to patients receiving treatment. A total of 56 papers were reviewed spanning a 6 year period from 11 different countries. 20% of patients received treatment through the MTB process. Of those that did not receive treatment the main reasons were no mutations identified (26%), no actionable mutations (22%) and clinical deterioration (15%). However, the data was often incomplete due to inconsistent reporting of MTBs with only 54% reporting on patients having no mutations, 48% reporting on presence of actionable mutations and 57% reporting on clinical deterioration. Patient attrition in MTBs is an issue which is very rarely alluded to in reporting, more transparent reporting is needed to understand barriers to treatment and integration of new technologies is required to process increasing omic and treatment data.

Retention in care and viral suppression after same-day ART initiation: One-year outcomes of the SLATE I and II individually randomized clinical trials in South Africa

Introduction: Same-day initiation (SDI) of antiretroviral therapy (ART) for HIV consistently increases ART uptake, but concerns remain about higher attrition from care after initiation. We analysed twelve-month retention in the SLATE SDI trials. Methods: SLATE I and SLATE II were individually randomized trials at public outpatient clinics in Johannesburg that enrolled patients not yet on ART and administered the SLATE I or II algorithm, which included a symptom self-report, medical history, brief physical examination, and readiness questionnaire, to assess eligibility for SDI. ART uptake and early retention have been reported. Using routine clinic records, we conducted a pooled analysis of retention in care and HIV viral suppression 14 months after study enrolment. Results and discussion: We enrolled 1,193 study participants (standard arms, n=599, 50%; intervention arms, n=594, 50%) and analysed by originally assigned groups. By 14 months after study enrolment (equivalent to 12-month retention in care), 50% of intervention arm patients and 46% of standard arm patients remained in care at the initiating site (crude risk difference 4% (95% confidence interval -1% to 10%; crude relative risk 1.10 (0.97-1.23), with similar viral suppression between arms. Observed attrition from care at site by 14 months was high in both study arms, but we found no evidence that the offer of SDI led to greater overall attrition or lower rates of viral suppression one year after starting ART and may have led to small improvements. Same-day initiation may have shifted some attrition from before to after dispensing of the first dose of medication. Conclusions: An offer of same-day initiation of ART, following a carefully designed protocol to identify patients who are eligible and ready to start treatment, is not inherently associated with an overall increase in patient attrition from care.

Optimizing therapy sequence to prevent patient attrition in EGFR mutation-positive advanced or metastatic NSCLC

Clinical trial and real-world data in non-small-cell lung cancer indicate that 10–60% of patients that progressed on first- or second-generation EGFR-targeting tyrosine kinase inhibitors (TKI) do not receive systemic second-line therapy. In our article, we discuss efficacy, safety and treatment duration with different EGFR-TKIs and stress the need for delivery of the most efficacious therapy in the first-line. We also provide our perspective on analysis of circulating tumor DNA and the role of EGFR-TKI in combined therapies. Finally, we review new therapeutic options to overcome resistance to EGFR-TKI. We believe that overall treatment duration and access to different medications in subsequent lines of therapy should be considered when planning the optimal treatment strategy.

Prior treatment status: impact on the efficacy and safety of teriflunomide in multiple sclerosis

Background In this pooled, post hoc analysis of a phase 2 trial and the phase 3 TEMSO, TOWER, and TENERE clinical trials, long-term efficacy and safety of teriflunomide were assessed in subgroups of patients with relapsing multiple sclerosis (MS) defined by prior treatment status. Methods Patients were classified according to their prior treatment status in the core and core plus extension periods. In the core period, patients were grouped according to treatment status at the start of the study: treatment naive (no prior disease-modifying therapy [DMT] or DMT > 2 years prior to randomization), previously treated with another DMT (DMT > 6 to ≤24 months prior to randomization), and recently treated with another DMT (DMT ≤6 months prior to randomization). In the core plus extension period, patients were re-baselined to the time of starting teriflunomide 14 mg and grouped according to prior treatment status at that time point. Efficacy endpoints included annualized relapse rate (ARR), probability of confirmed disability worsening (CDW) over 12 weeks, and Expanded Disability Status Scale (EDSS) score. The incidence of adverse events was also assessed. Results Most frequently received prior DMTs at baseline were glatiramer acetate and interferon beta-1a across treatment groups. Teriflunomide 14 mg significantly reduced ARR versus placebo in the core period, regardless of prior treatment status. In the core and extension periods, adjusted ARRs were low (0.193–0.284) in patients treated with teriflunomide 14 mg across all subgroups. Probability of CDW by Year 4 was similar across subgroups; by Year 5, the percentage of patients with 12-week CDW was similar in treatment-naive patients and patients recently treated with another DMT (33.9 and 33.7%, respectively). EDSS scores were stable over time in all prior-treatment subgroups. There were no new or unexpected safety signals. Limitations include selective bias due to patient attrition, variability in subgroup size, and lack of magnetic resonance imaging outcomes. Conclusions The efficacy and safety of teriflunomide 14 mg was similar in all patients with relapsing MS, regardless of prior treatment history. Trial registration Phase 2 trial core: NCT01487096; Phase 2 trial extension: NCT00228163; TEMSO core: NCT00134563; TEMSO extension: NCT00803049; TOWER: NCT00751881; TENERE: NCT00883337.