ISPOR 2025
See Research Differently
Fit-for-Purpose Data Sets
Real world evidence is an essential part of research, yet it can be expensive and time-consuming to ensure its accuracy and tailor it to your specific study.
The Challenge
There are stringent requirements that need to be met for real world data to be used in evidence packages. Research teams need to overcome quality issues, completeness issues, and challenges in interpretation and abstraction across sites.
The data needs to be reliable, valid, and relevant to a specific endpoint to be used as valid evidence in drug approval and reimbursement decision-making.
The typical real-world data sets in the market are built off structured EHR, claims, and registry data. As a result, they are broad but not deep.
This lack of depth leads pharma teams to observational studies or registry builds to gather this data. Both are costly and time intensive, and custom chart review studies historically take a very long time — whether you work with hospitals directly or data vendors.
The Opportunity
As we all know, EMRs were not designed for research, so clinical detail lives in unstructured forms across the patient record. Efforts to unlock this rich value have been hampered by varying levels of data quality, distinct EMR setups and differences in documentation practices across sites of care. As a result, real world data are inconsistent and lack transparency in how they were produced. Not exactly real world data.
What if you could leverage technology and clinical expertise to access this existing unstructured data across sites?
Q-Centrix is the trusted clinical data partner for over 1,200 hospitals and health systems across the US, driving data quality initiatives, better patient care, and performance improvement. Our work includes data curation for over 300 registries, medical societies, and regulatory agencies.
This same clinical data can allow research teams to unlock hidden value and plug data gaps in their existing evidence packages. We use our unique combination of behind-the-curtain access to US health system data and systematic abstraction processes driven by clinical expertise to help pharmaceutical companies generate Fit-for-Purpose data sets with the data elements and level of clinical depth needed to support evidence submissions.
This is quicker and more cost-effective than traditional methods of gathering this data, with similar levels of data quality. Q-Centrix’s data sets can stand alone or be used to supplement existing data and fill specific gaps.
Real World Evidence in Action
Case Study
Examining Patterns in Breast Cancer Patient Characteristics and Treatment Throughout the COVID-19 Pandemic
Patterns in breast cancer patient presentation and time to treatment throughout the totality of the COVID-19 pandemic have not been comprehensively investigated. The objective of this study is to aid in filling this gap.
Methods
28,048 breast cancer patients with a date of diagnosis occurring from January of 2019 to June of 2023 were extracted from the Q-Centrix Clinical Data Warehouse, a proprietary database of de-identified clinical data produced through expert-driven human abstraction. The subset sourced from the Clinical Data Warehouse includes information from 49 hospitals, health systems, and cancer centers nationwide.
- The time from January of 2019 to June of 2023 was analyzed by 6-month time periods.
- Chi-Square tests were utilized to compare differences in patient characteristics and treatment variables across the 9 time-periods. A p-value cutoff of <0.05 was considered significant.
- Post-hoc pairwise chi-square tests were run with a Bonferroni correction to control for multiple tests.
Findings
- 38.3% of all patients diagnosed from late 2019 to early 2021 received their first treatment within 30 days of diagnosis
- 24.6% of patients diagnosed from late 2021 to 2023 received treatment within 30 days of diagnosis
- 60.8% of patients diagnosed from 2019 to early 2021 received their first treatment 31+ days post diagnosis
- 70.7% of all patients diagnosed from late 2021 to 2023 received their first treatment 31+ days post diagnosis (p-value <0.001)
Conclusion
This study depicts a shift in time-to-treat in periods occurring from mid to late pandemic (late 2021 out to 2023) as compared to the pre- and early pandemic periods (2019 to the beginning of 2021). Patients received their first treatment an average of 1 week later in the latter portion of the pandemic than they did in the pre and early periods of the pandemic. Additional studies are needed to understand drivers of these dynamics.
This abstract was originally presented at the 2025 ISPOR Conference in Montreal.
Advancing Pharmaceutical Retrospective Research with Fit-for-Purpose Data Sets
In a recent Q-Centrix white paper, we explored the challenges of obtaining the data needed to conduct research and sharing insights from pharmaceutical and hospital researchers.
Data must often be gathered from multiple facilities, leading to differences in how the data is curated and organized.
A lot of academic medical centers are trying to build [data curation] capabilities in house. For us, to be honest, that can pose a challenge, where if everyone is doing their own thing, and it’s not a standard format, it’s a bit harder. One academic medical center may be doing it one way and another may be doing it another way.
— Pharmaceutical research director
Hospitals typically lack the necessary infrastructure or training to ensure data consistency.
We stopped agreeing to chart review studies. They don’t pay well enough to justify the amount of heartache. I have too much staff turnover to train someone on the data directives and software unique to each study.
— Hospital research vice president
There can be a lack of
clear data provenance.
People keep coming to us with AI/NLP-created datasets that are a total black box to us. How was the data trained? How can we speak to the accuracy? And more importantly, if there are no expert abstractors involved, why is it still so expensive?
— Pharmaceutical researcher