Can quality of life research learn from restaurant reviews?

𝐖𝐡𝐲 𝐢𝐬𝐧’𝐭 𝐇𝐞𝐚𝐥𝐭𝐡-𝐑𝐞𝐥𝐚𝐭𝐞𝐝 𝐐𝐮𝐚𝐥𝐢𝐭𝐲 𝐨𝐟 𝐋𝐢𝐟𝐞 (𝐇𝐑𝐐𝐨𝐋) 𝐮𝐬𝐞𝐝 𝐦𝐨𝐫𝐞 𝐨𝐟𝐭𝐞𝐧 𝐢𝐧 𝐜𝐚𝐧𝐜𝐞𝐫 𝐦𝐞𝐝𝐢𝐜𝐢𝐧𝐞 𝐚𝐬𝐬𝐞𝐬𝐬𝐦𝐞𝐧𝐭𝐬 (𝐇𝐓𝐀𝐬)?

Imagine you’re trying to book a restaurant and checking reviews, but each restaurant is rated on a different platform, with varying criteria and scoring systems. Some platforms send out their questionnaires only in the summer, favouring restaurants with outdoor seating (yet you’re unaware of this seasonal bias). Some sites collect reviews only during the dining experience, providing quick and up-to-date feedback, but missing input from those who may have felt unwell an hour later. Others may have too few reviews, making the feedback seem unreliable.  

Well, these situations are similar to how we currently handle health-related quality of life (HRQoL) research. Just as bad review sites wouldn't help you find the best food for your taste or the best value for your money, the current approach to HRQoL research doesn't truly support patients, healthcare providers, or payers in making informed decisions.

An example

Recently, we reviewed the HRQoL data from the eight phase III randomised controlled trials in metastatic hormone-sensitive prostate cancer (mHSPC) published between January 2015 and September 2024 (the pdf is available here). We identified significant methodological heterogeneity across the publications, including variations in HRQoL measures, individual HRQoL endpoints, data collection time points, and statistical methods used for Patient-Reported Outcome (PRO) analysis. HRQoL was reported using eight validated questionnaires, ranging from one to four instruments per study, in various combinations and the data was collected longitudinally at different time points and over different periods.

Additionally, issues such as lack of blinding, absence of a priori hypotheses, insufficient statistical power and challenges in handling missing data increased the risk of bias (see below).

Remarkably, the results of these trials were published with a delay of 4 months to 4 years after the primary clinical outcome publications.

Why so many scales and methods in HRQoL research?

Perhaps the most desired quality-of-life outcomes for patients are relief of symptoms, improvement in daily functioning and minimal or no side effects. However, measuring these are challenging due to factors such as individual variability, disease specificity, subjectivity of experiences and the complexity of measurement. This is the reason we have ended up with so many scales and methods to measure HRQoL. But if these data don’t contribute much to decision-making and the publication of results can often be delayed, why do we bother conducting this research?

Ticking the regulatory checkbox

In 1985, FDA’s decision to require survival and/or quality-of-life data in cancer trials drove research in this area, as it provided regulatory and commercial incentives. Years later in 2002, Prutkina and Feinstein write:

“Despite extensive growth in recent years, the field of "quality-of-life " appraisal still evokes debate about basic perception of the concept and is accompanied by a plethora of indexes for measurement. One prime reason for the problems is that the measurements have been transferred from two separate sources - medical health status indexes and social-science population indexes - neither of which was designed for appraising the particular personal distinctions of the way people feel about their own quality of life.”

They argued that “A person's quality-of-life is a state of mind, not a state of health, which is uniquely perceived by that person” and can’t be accurately measured unless patients are allowed to express their own views. The solution they proposed was to return to the traditional approach of directly asking patients to express how they feel.

Twenty-two years later, we are still struggling with a “plethora of indexes for measurement”.

A single platform, a straightforward scale and a broad range of experiences

Although restaurant and hotel reviews occasionally leave us disappointed, one could argue that the system still functions better than how we, as patients, evaluate therapies and share our experiences. The platforms sharing restaurant and hotel experiences gather data from a high volume of customers in real-world settings, capturing a diversity of opinions. They use clear scoring systems and make it possible to select places based on specific customer preferences (e.g., family-friendly or fine dining). As a result, selecting and comparing options becomes more straightforward and reliable.

Research on HRQoL data, is currently conducted over a limited timeframe and at specific timepoints, which may not reflect the real experiences. Other challenges include the controlled environments of clinical trials, small and sometimes unrepresentative patient groups, the time burden on patients and the complexities of comparing results across trials. As a result, the interpretation of results becomes difficult for regulators, healthcare providers and payers, as well as for patients who rely on these information to make informed decisions about their treatment.