Analyzing Specialty Drug Performance: Using Komodo’s Healthcare Map To Overcome Common Challenges
For Commercial Life Sciences teams, claims data can unlock powerful, patient-level insights to tease out new patients who are starting a medication, understand the duration of therapy, or pinpoint a potential new indication for label expansion. However, even the best datasets are susceptible to inaccuracies such as sampling gaps and biases, keying, and miscoding errors.
Specialty Drugs, Specialty Challenges
Specialty products are particularly prone to these challenges due to the prevalence of limited specialty pharmacy (SP) networks, manufacturer-required data blocking, and relatively low overall patient volumes. Because of constantly changing gaps in these data sources, analytics based solely on raw claims can produce misleading or incomplete results – giving manufacturers an obstructed view of patient journeys and drug adoption.
As an example, olaparib (Lynparza), a PARP inhibitor, is a notoriously challenging drug to track in claims data. After plateauing for several quarters, reported sales began to rise dramatically starting in mid-2017. What, exactly, contributed to this increase? A lot happened in 2017 and 2018. Originally offered as a capsule, olaparib transitioned to a tablet form with different dosing and pricing. It added a new SP to its network and enabled many hospitals and clinics to dispense. It also published data and received FDA approval for new indications in maintenance for recurrent ovarian and BRCA+ breast cancer.
In theory, patient-level data from administrative claims data can be used to disentangle the impact of each of these factors. Unfortunately, visibility into prescriptions from traditional claims sources like switches and clearinghouses varied significantly over this period. As a result, raw claim trends deviated meaningfully from known sources of truth, such as earnings reports of net revenue, calling into question any quantitative insights derived from raw claims counts.
As presented in a winning poster session during PMSA 2022, we built an analytics module that enables robust, longitudinal tracking of new-patient adoption of specialty drugs by indication. The module, which uses Komodo’s Healthcare Map and Sentinel application, can be automated to drive scalable insights for enterprise Life Sciences organizations.
The breadth and depth of the medical and pharmacy claims sources in the Healthcare Map provided broad visibility across geographies, payers, channels, and settings of care. The intersection of known payer-complete, provider-complete, and other sources available in our Healthcare Map allowed us to identify trend breaks and correct for gaps in visibility or coverage more accurately than sources that rely solely on homogenous datasets. We estimated the probability of observation of each prescription event and then used a Bayesian model to calculate the expected number of actual prescription events.
Enabling robust longitudinal tracking of new patient starts and duration of therapy by indication, line of therapy, or patient subsegment, coupled with advanced analytics, could equip Commercial teams with new insights to understand the key drivers of drug performance.
Crystal-Clear View of Performance
The resulting projection model gave a significantly more accurate view of olaparib’s performance. The model’s revenue predictions tracked within 2% of reported revenues for any given 12-month period over the past seven years. And the model’s estimates of new patient starts by indication per month allowed us to quantify the precise impact of each driver on product performance. Ultimately, these metrics also provide insight into how these companies can focus their efforts to ensure these critical therapies are reaching the patients who need them most.
Bringing together multiple disparate data sources for a more complete view of patient behavior can help Commercial teams identify gaps, measure performance, find the right targets, and bring therapies to the patients when and where they need them, faster.