Life sciences: Increasing speed-to-insight in pharma

Revenues in the worldwide pharmaceutical market are expected to reach $1.2 trillion by 2024, growing by more than 6% per year for the preceding 6 years, according to Evaluate. However, the percentage of revenue spent on R&D is predicted to decline from 21% to 17% during the same interval. One factor that may be causing this trend is increased efficiency in product development and testing.

Contract research organizations (CROs) now conduct the majority of both clinical trials and real-world studies, and they have invested heavily in technologies that allow greater efficiency. Another trend is increased emphasis on real-world studies, since insurance and government payers are seeking evidence of effectiveness as a condition for reimbursement.

In clinical trials, subjects are carefully selected based on specific characteristics and are randomly assigned to a treatment group, or sometimes to a placebo group. Measures relating to safety and effectiveness are obtained as part of the approval process. However, the populations are often relatively small, and the timeframe is limited. In contrast, real-world analysis typically involves post-approval, long-term studies in which the population reflects a more general group.

Many real-world studies include analyses of data from sources such as anonymized electronic medical records (EMR) and insurance claims, which are examined to provide information about medical outcomes and costs. In a typical data analytics study, the researcher receives a selection of data from a vendor who has aggregated it and seeks to answer specific questions that are posed by the manufacturer. The dataset is typically broader than the one needed for the study, and generally requires refinement to select the target population.

Quality data is critical

Evidera is a business within Pharmaceutical Product Development, LLC (PPD), a global CRO and provider of evidence-based solutions to demonstrate the real-world effectiveness, safety, and value of biopharmaceutical and biotechnology. The statistical analysis plan is written by Evidera scientists, and then presented to the analysts who carry it out. “Our first step is to conduct basic checks on the data for quality,” explained Katie Mercaldi, principal data analyst at Evidera. Since the data has often not been collected for research purposes (e.g., claims data from insurance companies), it may not be consistent or complete; therefore, a quality assurance step is essential. “We then search the data for certain characteristics to include or exclude in the final target cohort of patients.”

Once a patient having the condition or treatment of interest is identified and meets the study inclusion and exclusion criteria, the analysis can begin. “We look at many types of real-world issues, such as drug adherence and treatment patterns over time,” Mercaldi continued. “The analysis might be intended to detect adverse events and see what associations can be found, or whether treatment adherence is at the expected levels.”

Analytics tools

The primary analytical tool Mercaldi uses is SAS, which is used broadly across the healthcare industry. “SAS provides the flexibility needed to accomplish our objectives,” she noted. For example, she has written a series of macros to expedite the analyses. “One of my macros identifies the format of an analysis variable, and then SAS determines what the output will be. If it is a yes/no datapoint, then the output will be the label of the disease and the number of patients that have it. If it is a continuous variable, then the output will be a mean and standard deviation.”

The case of missing data is particularly prevalent in real-world data since the study environment is less controlled than that of clinical trials. Analyses that do not account for missing data can result in distorted results when calculating averages or totals. “I wrote a simple macro to identify cases in which datapoints were missing. Sometimes, missing data is not a problem, and just needs to be quantified and described. In other instances, it could signal an issue with the data, like missing age data in a study with age requirements,” Mercaldi commented. “The ability to readily identify records with missing data streamlines the analyses and allows us to produce and validate the results more quickly.”

Although some analytics tools have a simplified interface so that non-analysts can use them more readily, there is a trade-off in performance. “They work well for simpler analyses,” observed Mercaldi, “but can break down for the more complex ones, such as studies with complicated selection criteria or coding structures.” The key is to match the software product and user interface to the appropriate application.

CROs help speed time to market in other ways. “Small pharmaceutical companies may not have connections to physicians whose patients would meet enrollment criteria for a specific clinical trial,” said Venita DePuy, a statistical consultant and owner of Bowden Analytics. “This can delay the start of clinical studies. Enrollment time is critical for getting products to market, especially for patient bases with very specific characteristics.” Bowden Analytics provides quick turnaround consulting services for CROs and pharmaceutical companies that need ad hoc or interim studies during clinical trials.

Also a SAS user, DePuy receives data that originates in doctors’ offices and is consolidated into a curated database. “The main issues we address are typically whether there is a change in health from the baseline condition, or an increase in survival times,” DePuy stated. “SAS is a powerful program and we can do everything from analyze the study data, to calculate the visit window for scheduled study visits, to predict when physician offices have conducted study visits for their patients.”

In addition, the SAS interface with Microsoft Office allows export of data to Excel so that non-SAS users can view results in a familiar application. “I can export results to Excel with color coding that flags potentially erroneous data or overdue visits to the doctor,” commented DePuy, “which helps track important events in the study.” SAS-generated reports can be sent to program managers each day so they can be aware of progress in the clinical trial process and address any lapses.