Amy Bamford
May 22, 2023
The business objective is to increase timely and meaningful engagement with physicians regarding the 4th line of therapy/treatment option, Qinlock, indicated for the rare conditions of Gastrointestinal Stromal Tumor (GIST). An objective of this poster is to itemize and explain the patient journey business rules (see figure 3:Setting Time Windows / Patient Treatment Journey) applied to monthly Medicare data. By applying the framework to the closed claims source, results will produce situational awareness of treating providers with new patients and/or patients advancing line of therapy. The secondary goal is to quantify the gain of knowledge from integrating Medicare-based insights into commercial data strategy (see figure 6: Venn diagram).
The goal of this poster is to provide thoughts and insights on the challenges and opportunities of working to unlock data in the rare disease and/or late stage line of therapy treatment space that protects patient privacy while also empowering NPIs and patients alike.
Oncology patient journeys typically require switching between prescription lines-of-therapy requiring longitudinal data to understand the paths to wellness, timing of events, medications, and settings of care. Meanwhile manufacturers, payers, and specialty pharmacies can elect to block prescription data—when they do—events of the patient journey are no longer visible to researchers for care modeling and pharmaceutical sales managers for decision support. When a physician's timely access to knowledge and medication can have a life-saving impact, data blocks in “open” claims data give pharmaceutical sales managers unreliable therapy switch detection, and in some cases missing a physician’s data entirely. The problem is only compounded when the disease is rare (Fig 1).
This field resource allocation challenge and an opportunity for timely delivery of life-saving medications jointly motivate the line-of-therapy and alerting framework presented herein. The two part solution overcomes data block barriers and is proven to add previously unidentified healthcare providers by leveraging a “closed” claims database, Medicare fee-for-service claims. It also presents a set of time-based rules to help detect line-of-therapy (LoT) changes to power an alerts program for timely- physician engagement. Lastly, Medicare-based alerts must be integrated into an ongoing commercial/open claims alerts program.
Prior to combining Medicare and commercial data, the manufacturer recognized the critical importance of identifying National Provider Identifiers (NPIs) who are treating advancing patients' treatment lines with pinpoint accuracy and timeliness. In order to more accurately model the oncology/GIST patient’s treatment journey, researchers require closed payer data. Closed payer data refers to the claims from one payor or network, ensuring patient journey signal continuity. Open payer claims data refers to aggregated information from processors and switches.
When these two data sources are used in combination, manufacturers can achieve the most optimal results. This is especially important in oncology because patients often switch providers and treatment options, spanning time, settings, and at times, parts of the country. This manufacturer conducted research to identify other data sources that could be incorporated into the existing data assets to increase visibility of these providers and resolve the discontinuous patient journey conclusions drawn from open claims data sources. Medicare is the largest closed payor data source.
With CareSet’s access to CMS claims, the manufacturer receives monthly NPI-level aggregated alerts from Medicare’s Parts A, B, and D data, featuring a short two weeks lag time. For example, when a provider submits a claim on May 30, 2023, CareSet will see that claim on June 15, 2023.
In Figure 2, there are several “mock” patient journeys presented based on claims-based evidence. A dozen such mock journeys are typical, each having dozens of events. If an open claims data source was used, that patient’s journey could appear partial or discontinuous, making line-of-therapy-based physician engagement unreliable (Fig 2).
Patient indications were defined using ICD-diagnosis codes for GIST provided by the manufacturer along with several rules to ensure confirmed GIST patients are monitored for alerts. Two encounters for GIST, on separate days, were required for inclusion. All medications of interest require an ICD-based cross reference to confirm the GIST indication over a 1 year look back period.
Then, a set of patient-level longitudinal business rules outlined below are coded using SAS programming language. These rules assist in classification of the patient’s treatment journey, in terms of diagnosis and line-of-therapy status: NDx, fist line (1L) start, second line (2L) switch, third-line (3L) switch, etc.
Leveraging the physician NPI numbers available on these claims then drives the outputs of NPI arrays, incorporated by sales management in their alerts programming and improving engagement for HCPs with patients approaching a fourth-line (4L) option. The periodicity and mean inter-arrival rate of visits and treatment switches was measured previous to this implementation of time value findings (Fig 3).
1. Look-back clean period for new patient starts
2. Overlap Period to discern mono/combo treatments
3. Fall-off period for an agent in combo to fall-off
4. Treatment holiday - the time period where patients can abandon, achieve remission, or become watch-and-wait, such that the next treatment is considered first-line.
When treatment naive patients are detected to begin 1L, as detected by a treatment of interest, GIST diagnosis, and no prior treatment of interest in the prior 365 days, the managing physician's NPI number is passed to the field team.
Similarly, as treatments are added and subtracted from the patient’s regimen, additional NPI-level alerts are created. Each alert creates an engagement opportunity for the field team, and as the alerts reflect later line advancement, improving the probability the patient may require Qinlock in the 4L setting (fig 4).
Leveraging the population coverage differences between commercial and Medicare claims, the data sets are highly complementary, however a Medicare-based alert can sometimes reinforce a Medicare patient found in the open claims (Fig 5).
.jpeg)
The successful initiative used an innovative claims source and implemented a timely model for HCP-engagement. Opportunely, the Medicare data source identified 1,200+ previously unknown (see figure 6) Provider NPIs who treat patients with GIST, in addition to the NPIs covered by open claims data sources.
This workflow added data for all HCPs treating Medicare fee-for-service patients, including competitive brand data and providers in cancer treatment centers which were hidden in open claims by manufacturer or facility data blocks.
Secondly, the monthly alerts program improved the sales team's engagement of providers with patients advancing their line of therapy. In fact, the manufacturer found a secondary benefit potential for ‘following’ a commercial patient who has aged into traditional Medicare. Lastly, the manufacturer gained an understanding of the importance of CMS’s de-identification standards and its impact on HCP and patient activation. During the poster session, the client, Tom Boulay of Deciphera Pharmaceuticals, and Amy Bamford, BDD of CareSet, will share the final insights (Fig 7).
