Patient-reported outcome measures are standardised questionnaires that ask patients about aspects of their physical and mental health. They are known as self-report instruments because due to being completed by patients themselves—as distinct from tools that are physician-reported or nurse-reported. The answers that patients provide reflect their own subjective evaluation of their condition or difficulty.
Patient-reported outcome measures should assess conditions or difficulties that have been shown by research to be valid and measurable. A construct is an emotional, cognitive or physical human phenomenon—which may either be beneficial or harmful to the individual experiencing it. Examples of psychological constructs are anxiety, job-satisfaction and wellbeing—but we can also measure more tangible constructs such as physical pain and fatigue.
What types of PROMs are available?
There are two main varieties of PROMs that are commonly used in healthcare and research settings:
- Generic PROMs measure aspects of physical and mental health that are experienced by most patients in the general population. An example of a generic PROM is the Perceived Stress Scale (PSS-10)—which assesses cognitive and emotional stress in response to situations.
- Condition-specific PROMs ask questions that relate only so a specific health condition or a treatment that applies to that condition. For example, the European Organisation for Research and Treatment of Cancer, established a frequently-used PROM that measures quality of life in people experiencing cancer (EORTC QLQ‐30; Aaronson et al., 1993).
Can’t I just develop my own PROM?
Factor-analytic research is required before we can say that an instrument measures a construct. We can’t just create a questionnaire and say that it measures something. Rather, we need to prove that the items or questions that compose the tool—hang together in a highly-related and predictable way, and are tapping the same underlying outcome. In factor analysis, a large group of respondents or patients (typically N > 1000) is asked to complete an outcome measure.
Researchers then use sophisticated statistical procedures to identify similarities in how people respond to the items. Specifically, they look at the correlation between item responses. If the items are highly-correlated, it is probable that they measure the same construct.
Sometimes, there are clusters of items that correlate more closely with each other than with other item clusters. This might reveal that a higher-order construct is actually made up of several sub-constructs. For example, the construct of pain-catastrophising (i.e. catastrophic thoughts and feelings about pain)—has been shown to consist of rumination, magnification, and helplessness.
For this reason, each item of the Pain Catastrophising Scale (PCS) measures one of the three phenomena—forming three distinct subscales. But, the subscales must also correlate with each other—which is how we know that rumination, magnification and helplessness tap the same underlying construct of pain-catastrophising.
When determining the validity of an outcome measure, we also need to prove that peoples’ responses tend to correlate relatively positively with related constructs, and negatively with unrelated constructs. For example, we would expect that people’s responses to a PROM assessing happiness would correlate negatively with their responses on a PROM assessing depression. If so, we say that the happiness PROM has discriminant validity. According to some experts, a lack of attention to discriminant validity has led to a “proliferation of constructs across psychology sub-disciplines, with construct redundancy now rife” (Hodson, 2021).
Recent concerns have been raised about the proliferation of constructs across psychology sub-disciplines, with construct redundancy now rife.
Gordon Hodson, 2021
PROMs afford multiple benefits to healthcare organisations—which we’ll discuss comprehensively in another post. For now, I hope this post helps you form an understanding of what PROMS are, and how they are developed.
