Paper of the Week: 13th January 2020
This week’s blog is brought to you by: Dr Joe McManners
Full reference and title from the journal:
Biases distorting priority setting:
Health Policy, Volume 124, Issue 1. January 2020. Pages 52-60.
Link to paper click here
Authors conclusion:
Despite vast challenges and extensive efforts, the outcomes of practical priority setting are scarcely documented. Why is this so? This is the key question of this study. That is, why are the outcomes of priority setting so poorly documented, e.g. in reducing low-value care, when the principles, regulations, and tools for priority setting are fairly well developed?
The author looks briefly at some rational explanations for the discrepancy between theoretical efforts and practical outcomes in priority setting. Then looks to non-rational explanations to explain the extensive use of low-value care and direct our efforts to free resources for high-value care. He argues that ignoring these effects is a mistake in priority setting, and that revealing them is the first, but crucial, step towards making priority setting more aligned with its own aspirations, and addressing them.
3V bottom line:
In the face of a lot of effort and growing evidence, low value treatment and testing grows in cost and volume. ‘Rational’ explanations for this aren’t enough. Exploring human bias in decision making gives us a greater understanding on why we persist with low value care, at the expense of equity. Examples such as ‘identifiability’ with individual patients, ‘white elephants’ and ‘failure embarrassment’ should be made transparent to help us address this issue.
3VH – Implications for value:
It can sometimes be a mystery why we persistently in healthcare use testing and treatment that either doesn’t work, is potentially harmful, or at the least is low value – that is giving poor outcomes for the resource input.
There are many reasons for this, but the paper of the week explores why our human nature may inadvertently lead to this result, often without even realising.
The background to this is that the cost of healthcare rises greater than the rise in inflation. Breaking this down, this is driven by three factors:
- Changing population demographics, including ageing (9%)
- Inflation and income growth effect (gross domestic product, GDP; 29%)
- Increases in the intensity of clinical practice and innovation (62%)
The Office for Budget Responsibility (OBR) has shown that the increased intensity of clinical practice is the most important factor of the three.
Although some of the increase in the volume and intensity of clinical practice will be beneficial, it is likely that much of it will be of low value and cause waste.
The Organisation for Economic Cooperation and Development estimated that these increases in clinical practice result in at least 10% of waste and low value – resources that could be used in better ways for other individuals. One of the ways to address this growth has been the practice of ‘priority setting’, so that increases in health spending are spent on higher value care.
The author in our paper of the week argues that the frequent adoption of low value care is not as a result of excess, ie it is not that we have already adopted all high value care. We know that there is significant inequity in health systems, so that high value care is not accessible (or at least utilised) by all. Therefore using resources (money, time, workforce, carbon) on low value care reduces the opportunity for high value care and reduces equity.
A simple example of this is alluded to in the paper: With fixed resources, a clinician prioritising a patient in front of them to offer a MRI scan that has marginal benefit for the patient will (perhaps inadvertently) make it harder for the unseen patient who really needs the MRI scan.
The author suggests that there are rational explanations of why priority setting fails, but even if they were addressed there are other confounding explanations, which he characterises as biases.
Cognitive biases that effect decision making and resource allocation mainly level of individual clinician (micro), but also at meso (e.g. local system) and macro (e.g national) levels.
Priority setting attempts to address aspects of this. Generally this is approached on a ‘rational’ basis, looking at evidence and cost-effectiveness, to try to reduce low value care.
The author acknowledges that there are ‘rational’ reasons why this often fails (for example evidence changes), but describes ‘irrational’ reasons, namely biases that mean that those making decisions about resource use are subject to cognitive biases that mean their decisions don’t follow what would be expected rationally.
This is a useful way of explaining the situation that we see commonly, that despite being aware of low value care, it remains widespread and variable.
There is a summary of a number of these biases below, as they are particularly interesting.
In the example above, the significant increase in MRI scanning for non-specific back pain[i], against guidelines, considering these biases can be useful. Availability heuristics, identifiability affect, positive cognitive feedback, and extension biases are involved in the decision making.
The common argument is that by decision makers making ‘irrational’ decisions, they are potentially reducing resource for higher value care.
For us, this is important as if we are to pursue high value care and equity, we need to understand the reasons this doesn’t happen and pursue strategies to address this.
Biases
Identifiable and singularity affect, the patient in front of the clinician take priority over those not. So resources are potentially based more on demand rather than justice or equity. This also is counter to priority setting priorities such as ‘severity, effectiveness and efficiency’.
Rejection dislike, people don’t like taking things away from others. This is linked to feeling that to do so is to be ungenerous.
Failure embarrassment affect, it is natural to not want to admit that treatment we have been offering for years is of low value, or worse of no effect.
Prominence affect, that a dominant factor in a complex decision making process is used. This can be to simplify but can crowd out other factors important in priority setting.
Status quo affect, there is a tendency to prefer what is known and comfortable.
Endowment affect, people want more to give up something than they would pay to acquire it. Similar concept in health.
Loss aversion, feeling of wanting to preserve what you have.
Anticipated regret and ‘better safer than sorry’.
Risk aversion, feeling from clinicians that it is better to be criticised for doing something than for not doing something. Link to defensive medicine.
Availability heuristics, something is used as it is there; ‘scan because you can’.
Sacred values, and taboo trade offs, some treatments are seen as untouchable areas.
Progress bias, biases towards treatments seen as good progress.
Complexity bias, tendency to think that more complicated treatments are better than simpler.
Extension bias, thinking that more is better than less.
Asymmetry of risks and benefits, that risks underestimated and benefits overestimated.
Positive cognitive feedback loops, finding more on testing can lead to thinking that more is needed in the future.
Prestige bias, that some conditions are seen as being more ‘trendy’ than others.
Imperative of action, ‘something must be done’.
Competence affect, to maintain competence procedures/treatments need to be used, even if low value care.
White elephants, pressure groups or single issue groups pushing for expensive facilities.
‘Boys and toys’ affect, technologies are fun but this comes at an opportunity cost for other patients.
[i] https://www.thelancet.com/series/low-back-pain