The limitations of PROMs in decision-making
Last week we saw how PROMs or Patient Reported Outcome Measures can help us to understand the impact of a healthcare intervention or treatment on a patient’s quality of life. But could they also be used in an economic evaluation?
In this article, we’ll look at six of the limitations of using PROMs to support decisions about healthcare resource allocation.
1. PROMs typically provide multiple scores which are not useful in an economic evaluation
PROMs often have a large number of questions. For example, the ePAQ questionnaire that you looked at last week has approximately 120 questions. These questions cover many different ‘domains’ or areas of health and quality of life, and there will be multiple questions within each of these domains that cover similar aspects.
For example, the ePAQ has areas on urinary, bowel, vaginal and sexual problems and within the urinary area, there are 5 domains (pain, voiding, overactive bladder, stress incontinence and quality of life) which each have several questions. One advantage of a PROM such as this is that it provides a large amount of detailed information on each domain, and can provide a summary across different domains. This is extremely useful to the clinicians who use this measure to assess the health of their patients over time and to focus their attention on key symptoms during their consultation with the patient. However, can a measure such as ePAQ be used to inform decision-making that decides how to allocate health care resources?
Economic evaluation enables a comparison of the new and existing interventions by comparing across them in terms of the benefits they provide in relation to their costs. If a PROM such as ePAQ was used to measure the benefits of a new intervention, we may find that health improves for some symptoms, such as overactive bladder, but worsens for other symptoms, such as pain. Does that mean that health has improved, worsened or remained the same?
This is why PROMs that do not provide a single score are not typically used to measure the benefit in an economic evaluation.
2. PROMs typically assume all questions or domains are equally important
Most PROMs generate a score for each domain by equally weighting all questions within that domain and by equally weighting the response options to each question. For example, suppose we are generating a ‘mobility’ score using responses to these two questions:
Q1. Ability to walk 5 km
a. I have no problems walking 5 km
b. I have some problems walking 5 km
c. I am unable to walk 5 km
Q2. Ability to walk 100 metres
a. I have no problems walking 100 metres
b. I have some problems walking 100 metres
c. I am unable to walk 100 metres
Producing a score that gave equal weighting would mean that being able to walk 5 kilometres is as important as walking 100 metres. However we might argue that the ability to walk 100 metres is more important than the ability to walk 5 kilometres. For example, we could easily drive, catch a bus or a taxi to take us a distance of 5 kilometres – but we might need to walk 100 metres to get to the car, bus or taxi. Also, they are both measuring ability to walk, and instead we might want to have a single question that can represent the full range of severity in mobility – for example, ability to walk with different distances as the response options - rather than these two questions that measure different ranges of severity for a set distance. In the same way, PROMs assume that all domains are equally important, for example pain and mental health, which you may not agree with.
3. PROMs typically assume that the difference in severity across all response options is equal
Most PROMs generate a score for each domain by equally weighting the response options to each question. In the example above this would mean weighting the response options like this:
|Q1. Ability to walk 5 kilometres||Score|
|I have no problems walking 5 kilometres||1|
|I have some problems walking 5 kilometres||2|
|I am unable to walk 5 kilometres||3|
|Q2. Ability to walk 100 metres||Score|
|I have no problems walking 100 metres||1|
|I have some problems walking 100 metres||2|
|I am unable to walk 100 metres||3|
The increase in score is the same for moving from no problems to some problems as moving from some problems to being unable to walk. However we may think that moving from some problems to being unable to walk represents a larger impact on severity than moving from no problems to some problems.
4. PROMs do not typically provide a score that reflects how important we think their impact is on our health
The scores generated by a PROM will depend on how many questions and domains are included in the PROM, and what these are. However, for economic evaluation we are interested in the overall health benefit from an intervention and how important that is. Simply put, a patient may have extreme problems with constipation but this may have a small impact on their overall health, whereas they may have moderate pain that has a large impact on their overall health. Only by collecting preferences about different descriptions i.e. how good or how bad people think different descriptions are, can we determine the overall health benefit from an intervention.
5. PROMS focus on health related quality of life
PROMs typically focus on health related quality of life but conditions and interventions can have an impact on both the quality of life and the quantity of life. In other words, both morbidity and mortality are important outcomes in health.
6. There are many different PROMs all with very different content
Finally, it is a huge advantage for research that there are many different PROMs with very different content, as this allows us to use the ones most suitable for the patient group and our research question. However, for economic evaluation we are deciding how to allocate healthcare resources across many different interventions for many different patient groups. The use of different PROMs for different interventions and different patient groups means that we do not have a single measure of benefit for use in economic evaluation, which makes comparability across evaluations nearly impossible.
© The University of Sheffield