Section 1: Introduction: Causes of “Overtreatment” from both the individual and societal level.
My sense is that each section of the Commission will have to present a working definition for what they see as ‘overtreatment’. Defining and then quantifying overtreatment will be the major focus of the Economics Component. Although there are likely many definitions of what constitutes overtreatment, from an economics perspective it is defined as treatment where the marginal costs exceed the marginal benefits. Due to generous insurance coverage, healthcare is rife with overtreatment (termed dead weight loss by economists) and many economists have made careers out of measuring the extent of it. The most famous study is the RAND Health Insurance Experiment in the US from the late 1970’s which found that when consumers were required to finance their own health care, they reduced spending by about 30%, suggesting that at least 30% of healthcare would fit into the definition of overtreatment.
There are many reasons to suspect that this figure represents a lower bound. Other drivers of overtreatment include biases among patients, families and doctors that push people toward low value treatments where the costs likely exceed the benefits, and even more so for patients with life limiting conditions. Optimism bias/hope is the primary bias but there are many others. Even in the absence of bias, families push for low value treatments to maintain (often false) hope and to avoid regret and doctors do it to avoid having tough conversations with patients. Institutional factors (e.g., profit, greed) also lead some doctors and other stakeholders to push for treatments over supportive care.
Section 1 will expand on all of the above issues with a focus on likely causes of overtreatment at EOL from the individual, family, physician and societal perspective.
Section 2 Overtreatment from the Societal Perspective
Section 2a (A Stated Preference Survey with Reed Johnson, Semra Ozdemir, Juan Marcos Gonzales, Eric Finkelstein and Richard Smith as co-authors, others TBD)
As noted above, a primary cause of overtreatment at EOL is generous insurance coverage in UHC countries. However, as the level of coverage is a collective decision by government (elected by taxpayers), in theory that level of overtreatment is acceptable to constituents and preferred to a situation with less coverage and less health equity. However, this assumes that governments are appropriately executing the preferences of their constituents. That will be the focus of this section. We will address the following questions about overtreatment from the taxpayer perspective via a stated preference survey in multiple higher and lower income countries (countries TBD and based on available resources).
- Overtreatment RQ1: Is the percent of government budget allocated to health services too high relative to other sectors gov. could spend money on.
- In essence we will ask respondents to split out the government budget pie across various sectors that governments spend money on.
- We would not anchor respondents to a base case but we would show them the range of gov. spending (%’s) allocated to various sectors across different countries (e.g. health care ranges between x and y %, military between a and b %, education between c and d %,…)
- The optimal allocation (%) dedicated to health services from the taxpayer perspective will then be compared to the actual gov. % within a country to provide a measure of overtreatment on the health sector relative to other sectors. We will then compare within and across countries by key strata (e.g., UHC,…)
- Note – this gets at the optimal allocation but not the optimal size of the pie. That is something we may or may not be able to quantify. TBD.
- Overtreatment RQ2: Is the percent of the government Health Services budget allocated to non-palliative EOL treatments (i.e., QOL improving and/or moderately life extending treatments) too high from the tax payer perspective relative to other types of health services government could spend money on? This is the key question for the Lancet Commission
- We will replicate the exercise in RQ1 but the pie will be limited to different types of health spending that govs. could spend money on, including non-palliative EOL treatments, curative, palliative (not life extending), chronic disease management, mental health,…
- Data permitting we will again compare the optimal allocation (%) with actual expenditures to quantify overtreatment of non-palliative EOL treatments from the taxpayer perspective.
- Overtreatment RQ3: How should non-palliative EOL $ be split among the various types of QOL improving and/or moderately life extending EOL treatments available.
- Here we will most likely use a discrete choice experiment to quantify trade-offs between types of EOL care that improve quality and/or quantify of life to varying degrees.
- This question allows for getting at the types of EOL care that taxpayers see as low value and the extent to which they would trade off one type of care for another.
- One way to operationalize this would be the person-tradeoff approach that WHO uses. This can be done by splitting EOL care into the following categories (as an example):
- Extend life by: 0 months, 3 months, 6 months, 12 months, or 18 months
- Improve quality of daily life until the dying period by: no improvement, moderate improvement, large improvement
- Number of people helped.
- We will then show participants combinations of the above and ask them to choose which they prefer. This allows for valuing various combinations of quality and quantity of life in terms of number of people helped. We may find that there is very little value placed on non-life extending care regardless of its impact on quality, for example.
Note that the plan for the above is to survey taxpayers. Resources permitting it may also be of interest to survey other key stakeholders, including doctors, managers, nurses, politicians, etc although getting them to take surveys is very tough.
Section 2b (Evidence based on existing data, with Brett Doble, others TBD)
An alternative approach to quantifying overtreatment at EOL. If countries inherently value EOL care similar to how they value other types of care, then presumably this care would be approved and funded at the same incremental cost effectiveness ratio as non-EOL care (ICER). If so, then under certain assumptions, we would expect that the percentage of costs that go toward EOL care in a year should be roughly equal to the percentage of people who die. However, this is unlikely to be the case. In UK, for example, annual spending on EOL care is roughly 15% and the percent who die is below 1%. French et al do not provide mortality estimates but present statistics on % of aggregate annual expenditures that go to EOL in last year of life with estimates ranging from 8.45% (US) to 11.20% (Taiwan). Given that death rates are well below these levels, this provides strong evidence that we are spending money on EOL care for interventions with cost-effectiveness ratios many times that for non-EOL care and likely above established thresholds for cost-effectiveness. Ideally we would present these statistics for as many countries as we can find and make the point that this is true across all developed countries but unlikely to be true in developing countries (note: I am not sure how much data exists so this is something we would need to look into but the idea is to present existing data in an interesting and new way to highlight money is not being well spent). We would supplement this data with other economic markers suggesting low value of EOL care based on what is available in the literature.
Note that one caveat to the above recently pointed out by Einav et al is that these determinations are made ex post. Ex ante some deaths were meant to be prevented not just prolonged. Our focus is on those treatments where the goal is only to moderately extend life and/ or improve quality of life but not meant to be curative. However, it is unlikely that we can identify what percentage of deaths falls into this category although it is likely to be the majority for those who die of cancers and other chronic diseases
- French, E., McCauley, J., Aragon, M., Bakx, P., Chalkley, M., Chen, S.H., … & Kelly, E.(2017b). Data from the U.S. and eight other developed countries show that end-of-life medical spending is lower than previously reported. Health Affairs, 36(7), 1211-1217.
- Predictive modeling of U.S. health care spending in late life. EinavL, Finkelstein A, Mullainathan S, Obermeyer Z. Science. 2018 Jun 29;360(6396):1462-1465. doi: 10.1126/science.aar5045.
Section 3: Summary and Recommendations
Recommendations for reducing overtreatment at the individual and societal level will come from what we find above, Felicia’s earlier Commission work, and work from the other parts of this Commission.