The Comparative Health Outcomes in Immune- Mediated Diseases - PowerPoint PPT Presentation

The Comparative Health Outcomes in Immune- Mediated Diseases Collaborative (CHOICE) Study Jeffrey Curtis, MD, MS, MPH Professor of Medicine @RADoctor Shilpa Venkatachalam, PhD Associate Director, Patient- Centered Research GHLF

Jeffrey Cur Curtis, M , MD, , MS, M , MPH: Consulting fees and contracted research with AbbVie, Amgen, Bristol-Myers Squibb, Corrona, Eli Lily, Janssen, Maraid, Phizer, Regeneron, Roche, and UCB. Shilpa a Venkatac achal alam am, P PhD: I can confirm that I do not receive any direct funds from industry related to RA, nor do I in any capacity bear relationship with industry, including pharmaceutical companies related to RA that may pose as a conflict of interest in my capacity researcher for the CHOICE study. I am currently employed by the Global Healthy Living Foundation as the Associate Director, Patient Centered Research. GHLF receives grants and sponsorships from pharmaceutical manufacturers as well as other private foundations. A full list of GHLF funders is publicly available here: https://www.ghlf.org/our-partners/ None of these partners, however, have the potential to bias or appear to bias my involvement in this project. Any possible conflict of interest that might arise will be reported promptly by GHLF.

Resea earcher P Per erspecti ective: C CHOICE S Study

Backgroun und: d: Need to make challenging decisions amongst a variety of treatment options for autoimmune and inflammatory conditions given widely variable safety profiles; limited data on medication effectiveness from patients’ perspective Aim 1 m 1: Safety of immunomodulatory medications (e.g. biologics), researcher perspective Aim 2 m 2: Comparative effectiveness of immunomodulatory medications, patient perspective Data i infrastructure  EHR data from 3 CDRNs  Patient data from 5 PPRNs  Health plan claims Recruitment ( (Aim 2 Sub ub-AIM) M)  CDRNs recruit patients to PPRNs  PPRNs recruit patients to Study

Example le o of v f varia riabili lity in CDM d data provided • Low percent = GOOD • Includes zero refills for autoimmune prescription refills

Variability i in Sidecar Data p provided

AIM 1 1 Safety R ty Results ts f from om C CDRN S Sites De Def 0) 0) no requirement for any prior data; De Def 1) 1) >6 months (m) from first medical inpatient or outpatient medical encounter of any type; De Def 2) 2) >6m from first prescription [Rx] for any medication; De Def 3) 3) >6m from first medical encounter for disease indication (e.g. RA, vasculitis); De Def 4) 4) >6 m from first prescription for any disease-specific rheumatologic therapy

Total patient data received from CDRNs N = 183,372 AIM 1 1 Data Patients with some autoimmune disease that c could N= 170,032 be l be link nked t to Patients with RA, PsA, PSO, IBD, AS, Vasculitis or JIA diagnosis code CMS D Data N = 173,545 Patients with any medication information (including parenteral medications) N = 140,838 Patients with any RA, PsA, PSO, IBD, AS, Vasculitis or JIA medications N = 63,885 Patients classified as having RA, PsA, PSO, IBD, AS, Vasculitis or JIA after using a Rx/Px for these diseases, but no additional drugs for these diseases afterward N = 31,878 After excluding for HIV/Cancer/SLE/Organ Transplant/autoimmune diagnosis before 6 months N = 37,453 After excluding for vasculitis (except for the vasculitis cohort) N = 37,393 Patients linked to CMS data N = 1,956

5,637 Patients with Medicare (Medicare advantage) 5,559 Patients with date of birth (DOB) ArthritisPower patients (AIM 5,545 Patients with sex 2) who could be linked to 5,544 Patients with zip code CMS data 2,352 Patients uniquely linked based on DOB, sex and zip code

AIM 2 Patient Reported Outcomes (PROs) Example from PARTNERS data analyses Aimed to compare baseline PROs to 6 months after newly starting a medication for treatment of JIA Inclusion criteria: • PRO measured with 30 days prior or 7 days after newly starting medication (methotrexate or biologic) • Subsequent PRO measured between 3 and 8 months after newly starting medication Outcome: • PROMIS measures of Pain interference, Physical function, Fatigue • Patient/Parent Global Assessment of Disease Activity • If newly started medication was discontinued prior to 6 month PRO measurement, then most recent score (including baseline score) was assessed.

MTX TNFi MTX+TNFi monotherapy monotherapy combination % RF+ Polyarthritis 12% 10% 25% % Spondyloarthritis 18% 50% 23% Mean disease duration 222 580 110 (days) % High disease activity 78% 79% 87% Impor ortant D t Differ eren ences ces in B Baseline e Characteristi tics cs Initi tial analyses show t that h t high gh d disea ease a acti tivity ty i is strongly a associated wi with s subsequent t improvement i t in PRO ROs.

MTX TNFI MTX+TNFI MONOTHERAPY MONOTHERAPY COMBINATION Number of patients 176 42 40 Baseline Score 55.41 / 55.70 57.05 / 58.00 57.95 / 59.00 Mean/Median Mean Delta -4.11 -2.75 -7.51 Median Delta -2.70 -0.50 -5.75 Changes es i in PROMIS P Pain I Inter erfer erence ce

MTX monotherapy TNFi monotherapy MTX+TNFi combination Number of patients 193 31 39 Baseline Score 39.21 / 37.90 37.78 / 36.90 35.27 / 34.20 Mean/Median Mean Delta +7.37 +6.69 +10.20 Median Delta +6.00 +4.90 +6.90 Chan anges i s in n PROMIS P Physi sical F Function

• Clinically relevant improvements in PROs were observed following initiation of new medications PARTNERS AIM 2 • There appear to be differences in improvement across different medication choices, but Summary to Date important baseline differences in patients’ characteristics (especially disease activity and possibly disease phenotype) necessitate adjusted analyses

Ad Admini nistratively • SMART IRB functionality was limited • Variability in SMART IRB capacity among CDRN/PPRN sites • Local IRB approvals still required by sites even when using SMART IRB • Template DUA not accepted “as is” for all sites; changes requested AFTER execution • Uncertainties about DUA permissions to share LDS (vs. de- identified) • Contractual processes and governance structures varied by CDRN and by individual site • Contractual/IRB dependencies – i.e. Human subjects protocol cannot be submitted to IRB without contract number, but contract cannot be routed without IRB approval; DUA requires IRB approval

AIM 1 • Should involve ALL possible stakeholders early in the process (i.e. statisticians/programmers were not oriented to the protocol when it was time to extract the data) • Variability in CDRN data that was available (both CMD and sidecar) • CDRNs had varying levels of standardization to the CDM, with some having no centralization. • Missing 45-100 % of refill data on autoimmune diseases. • Mother-baby linkage data was not available at four of the five data marts because (1) IRB nightmare to get baby IDs, (2) there was a limited number of female RA patients of child-bearing age who had babies at the data mart and (3) it was just not available. • PCORnet CDM was inadequate for representation of the data • QA/QC was difficult between CDM versions (data was extracted by the data marts, but while waiting for DUA execution, the data marts updated their CDM versions) • Necessary to have a physician champion at each data mart

AIM 2 • No harmonization in PRO data collection, frequency and method of collection by PPRNs • Variability in flexibility of PPRNs to modify existing PRO collection versus using concurrently collected data for other studies • Establishing the baseline anchor not possible in all PPRNs - Added two additional outcomes to compensate • Standardized CDRN to PPRN recruitment letters proved an impossibility • Variability in CDRN recruitment capacity • One data mart’s patient portal team requested changes to the vetted and IRB approved recruitment letters requiring a 4-month delay • One data mart waited 3.5 months for IRB approval related to messaging patients through the patient portal • One data mart lacked the capacity and had to build the infrastructure • One data mart was unable to engage in active messaging (could use passive messaging) • PPRNs were split over whether to invite members to JOIN the CHOICE study vs. ask them to provide more (or regular) data without actually joining the study

Stakeh ehol older er Perspect ective

What do we mean by stakeholder (patient) perspective? Patients are the only o ones who can tell researchers how well or not well a treatment works for them, how to best contact them, and how often they want to provide data. Why include patients? • Patient Reported Outcomes (PROs) can only come from patients. • This data, from thousands of patients, is then turned into information that physicians and other patients use to make decisions about treatments. • But, how can researchers feasibly connect with and engage thousands of patients

Patien ent P Power ered ed R Resea esearch h Net etworks ( s (PPRNs)

User opens app and is prompted through a ‘wizard-like’ process to participate in a study via a walk-through tailored to their cohort providing an enhanced user experience.