Lab Data Bottlenecks & Breakthroughs: Inside Charlie Kreilick’s 20 Year Playbook

Charlie Kreilick is a healthcare data and analytics leader with over 20 years of life sciences experience.

1. Can you share a bit about your background, in particular with working with lab data, and how that experience has shaped your perspective on leading RWE teams?
I have led Global Data Analytics organizations in the pharmaceutical industry for more than 20 years. The primary focus has been in the Health Economics and Outcomes Research (HEOR) and the Real World Evidence (RWE) space. Both areas rely heavily on Real World Data (RWD). RWD is primarily aggregated across many data sources by third party vendors. These vendors typically do not alter the data as long as it meets the data structure requirements. This often leads to missing and/or erroneous data. It has been my responsibility to explore all options to curate the data, as best we can, to ensure the data is research-ready and fit for purpose and that we meet our timelines with the highest level of scientific rigor.

2. In your experience, what are the biggest challenges lab data faces when it comes to integrating into analytical or research workflows?
An integral part of RWD is the Laboratory component. These data contain information on lab tests ordered for patients and their subsequent results. Lab tests are identified by LOINC codes. These codes are often missing or are coded in error resulting in missed opportunities. The challenge here is that the analysts have no way of knowing how to impute the missing codes or correct the erroneous ones.

Test results contain information that suffers from data quality as well. Units of measurement are often found to vary from the standard. For example, a test that has a standard measurement of MG can also have several other units (i.e. mmol). The challenge here is to be able to standardize to one single unit of measure and then recalculate the actual test result for those that were changed. This is a labor-intensive process to an analytics organization that can add an additional couple of weeks to a research project.

3. Could you give an example of a project or initiative where improved lab data quality significantly impacted research timelines or outcomes?

There was an RWE research project that used certain LOINC codes as part of the inclusion criteria for the cohort definition. Lab tests and results were also used to determine the efficacy of the product. The database that we used had a high level of missingness to the LOINC codes as well as a large distribution of units of measurement that needed to be normalized to the standard. Normally this would take our data analysts weeks to fix manually. Using Lab data that we had cleaned and standardized not only saved the group weeks of tedious work, which kept us on schedule, but it also was able to identify more patients in the cohort. This cohort was already small in size. Our group had no experience in the imputation of  missing LOINC codes so those patients would never have been found. Finally, we had a higher level of confidence in our outcome measures because of the improved quality of identifying/correcting missing LOINCS and the standardization of the results.

4. Are there certain clinical research areas where lab data is more relevant than others?

Lab data is important in HEOR/RWE research in a couple of ways. The first is in cohort identification. A cohort can use certain Lab tests as an inclusion criteria for entry. The second is in outcomes. Lab tests are used to help define certain outcomes of interest. Lab test results can be used for outcomes measurement of interest to show efficacy of a company’s product. 

5. Looking ahead, what emerging trends or technologies do you see shaping the future of lab data and how can solutions like Cornerstone AI’s lab product help pharma leaders stay ahead of the curve?
Lab data will inevitably be used in pharmaceutical research one way or the other. A company may not have a product in the market now that uses Lab data as part of a cohort definition or as an outcome measure, but with all the pipeline products, mergers, acquisitions, or indication adds/switches, things can change in an instance. Having data that is ready for research is key to getting the message to market first. Being proactive with the preparation of your data will help achieve this as opposed to being reactive and waiting months for procurement, legal, and budgets to approve before you can even start the harmonization of your data.