menu
future

Gathering real world data for new therapies is a vital part of demonstrating value to payers and providers in the current healthcare landscape. As they enter the market, emerging therapies and devices can be difficult to study because the precise characteristics of the patient population that will receive the drug are unknown. These products can be subject to rapid changes in prescribing or use over short periods due to recent market launch, approval of a new indication, competitive displacement, or safety warnings. In safety and effectiveness studies of such products, these rapid changes can lead to analytic challenges and opportunities.

While trends in prescribing can be challenging to track, they offer analytic opportunities to study how products disseminate through the market and insights into changes in utilization as physicians and patients gain experience with new therapies. If a product experiences rapid changes in the midst of a multi-year study, for example, researchers need to examine the impact of dissemination over time and consider how it may impact results.

My colleague, Mugdha Gokhale and I recently wrote a paper on this topic for Current Epidemiology Reports, entitled “Toward an Understanding of the Challenges and Opportunities when Studying Emerging Therapies.” We discuss how to identify dynamic patterns in dissemination or use, and how to account for and potentially take advantage of these patterns in safety and effectiveness, and includes several case study examples.

Prescribing over time

Studies of drugs or devices that experience profound changes in prescribing or use over short period of time warrant additional examination and, potentially, tailored methods. When analyzing the impact of these products in real world settings, it is important to consider what dictates these shifts and why they occur. Initially, it may be driven by new-to-market status or approval for a new indication, though rapid changes in prescribing can also be due to competitive displacement or safety signals that curb or halt use.

Such rapid changes need to be accounted for in the selection of study cohorts and data assessment approaches, which isn’t always easy. Analytic methods to account for these patterns in safety and effectiveness studies do not have a one size fits all solution; rather, they require individual investigation and contemplation of how best to incorporate information on dissemination and use.

For example, if in the middle of a two-year study, a safety warning dramatically shifts prescribing patterns, researchers need account for that when assessing the data, rather than averaging the results across all users over the entire time period. Similarly, the cost of a new drug may directly impact who uses it in the first year, but once payers begin to support that product, average income of users will shift, which may change the need for adjustment by socioeconomic status.

This underscores the importance of examining the impact of calendar time throughout studies of all treatments, and particularly those that are candidates for dynamic dissemination. Acknowledging the impact of time on these studies has become an even more relevant issue with the increase in post-approval effectiveness studies and strong practices for safety monitoring of existing treatments. Comparative studies in particular may be more impacted by time trends, since the clinical balance between treatments may change if there is a safety signal, or the introduction of a second treatment options causes rapid changes in use.

Early studies of emerging therapies and devices also offer a chance to examine causal relationships in variable patient populations that is similar to the insight captured in a randomized study. Being able to examine safety and effectiveness in different and shifting patient populations helps researchers identify anomalies that might otherwise be missed, which can give them greater confidence in their results. One of the examples we explore in the paper is a retrospective analysis of Medicare enrollees using two different oral second-line antidiabetic drugs between 2008 and 2013. The two initiation curves crossed in 2010 around the time of FDA warnings about the safety of one of the drugs. Due to such dynamic patterns of initiation over a relatively short period of time, the authors were able to examine cardiovascular outcomes comparing use of the two drugs in restricted populations, which revealed that in spite of the dynamic treatment patterns, no difference was found in the characteristics of patients selected in early versus later years of this restricted population.

The paper includes several other examples, and offers researchers strategies for studying dynamic therapies, and advice on how to use advanced methods such as calendar time specific propensity scores and instrumental variable studies to account for bias and better understand the impacts of treatment dissemination.

There are a number of methods available for understanding, accounting for and taking advantage of these dissemination patterns in study design and analysis. While excellent literature exists individually on different methodologic options, we recommend diligence to this process as well as provide a tactical explanation of approach. Ultimately, rapidly disseminating therapies require a manual look and thoughtful consideration of patterns in data, as well as knowledge of clinical use and patient population. When researchers take the time to assess how shifting patterns impact their research, and to take advantage of these shifts, they will deliver better, more reliable data to support their research objectives. 

Topics in this blog post: Biopharma, Evidence, Registries, Population Health