OR WAIT 15 SECS
In this article, we will analyze themes from asthma patient conversations via HealthUnlocked, an online healthcare social network, and compare these themes to asthma clinical trial endpoints from large pharma studies.
John Whyte, Director of Professional Affairs and Stakeholder Engagement at the FDA, suggested that while some in the industry are selecting meaningful clinical trial endpoints, the biopharmaceutical industry can often do better in selecting more optimized outcomes by choosing clinical trial endpoints that are clinically meaningful to patients. He believes regulators and sponsors need to improve listening with patients and not simply talking at them.
There is a potential approach by listening with patients in aggregate through the use of big data research. In this article, we will analyze themes from asthma patient conversations via HealthUnlocked, an online healthcare social network, and compare these themes to asthma clinical trial endpoints from large pharma studies.
HealthUnlocked is the third largest health website in the UK and in the top 20 globally. With over 700,000 members and approximately 40 million users per year, the website increases patient engagement and contributes to improved health outcomes, through use of its online health and wellbeing forums/communities. Users gain access to useful information and relevant services, seek advice, support, and mentorship from people like them, are empowered to make better healthcare decisions, and help advance medical research and understanding.
Data and Methodology
The objective of this research is to identify common themes discussed among patients from HealthUnlocked’s asthma community (clinically meaningful outcomes to patients), and to compare those outcomes to clinical trial outcomes from 15 large pharma studies attained from www.clinicaltrials.gov.
We analyzed approximately 152,000 unique conversations between asthma patients from February 2006 through June 2017. HealthUnlocked used its own natural language processing (NLP) algorithms to create tags for each discussion (defining specific topics of discussion). We analyzed these tags from approximately 152,000 unique conversations between patients (in the US and the UK in aggregate) using a Latent Dirichlet Allocation (LDA) algorithm, a machine learning NLP algorithm that predicts subsequent words in a conversation, and categorizes conversations in aggregated topics on an inter-topic map, as illustrated in Figure 1. Figure 1 delineates that the LDA algorithm has derived five main topics from the conversations; in this article, we will analyze the top three topics.
Details of Topics
Figure 2 demonstrates that the most common topic of discussion with asthma patients includes lung and respiratory infections, whereas Figure 3 (the second most common topic) exhibits that allergies, medications, and walking are topics of interest between patients. Figure 4 (the third most common topic) indicates that patients are discussing allergy related topics including rhinitis, and hay fever.
Best Fit Conversations
In order to better understand this analysis, we dug deeper into the conversations by requesting the algorithm to find the most fit conversation for each topic of discussion. Table 1 demonstrates these conversations.
Are Pharma Endpoints Clinically Meaningful to Patients?
In order to answer this question, we conducted a comparative analysis by evaluating 15 big Pharma asthma studies from www.clinicaltrials.gov, aggregating those studies’ endpoints, and comparing those endpoints to the patient reported endpoints from the aforementioned analysis. Table 2 suggests that the biopharmaceutical industry is taking a general and broad approach towards addressing endpoints, whereas patients are expressing specifics with regards to what endpoints they consider clinically meaningful. To elaborate, the ACQ-7 questionnaire focuses on asthma exacerbations, general activities impacted by asthma, shortness of breath, and use of inhalers. Alternatively, patients are more concerned with the number of infections, the rate of bronchitis/bronchiectasis, the amount of allergic exacerbations, and improvement in walking.
*Pharma outcomes analysis is based on reviewing study endpoints from 15 big pharma Phase II & III studies on www.clinicaltrials.gov. Companies include GSK, Sanofi, Novartis, Genentech/Roche, Merck Sharp, and AstraZeneca.
Sponsors Can Use Aggregated Technology to Optimize Study Endpoint Design
Naturally, the use of aggregated data analysis does not necessarily provide the rigor and validation techniques to determine study endpoints as recommended by regulatory authorities; however, use of such tools can be extremely valuable when developing research strategies with patients during study design. Moreover, sponsors can leverage such assessments in order to better understand their competitive advantages in the marketplace, when designing their studies.