This study shows that eliminating cancer at the current exchange rate between money and health would increase total health expenditures in Germany 3.07-fold or by 207% in the base-case analysis and 2.35-fold accounting for generic/biosimilar entry. The underlying gain in life expectancy from cancer elimination is in line with the results of other studies. For example, the gain for female and male newborns in the Netherlands was reported to be 3.6 and 4.1 years, respectively, based on data from 2009 , whereas in the U.S. the gain for newborns was estimated to lie between 2 and 3 years in the period between 2001 and 2008 .
Based on the gedankenexperiment the percentage of income spent on SHI in Germany would grow from currently 15.7% (which includes an average supplementary premium of 1.1% ) to 37% even considering generic/biosimilar entry. Disregarding the macroeconomic implications of such labor cost increase (e.g., in terms of competitiveness of German goods and products in the international market), the question appears whether the German population would support the necessary drastic reduction of non-health consumption. Also, the reduction of non-health consumption could reduce the survival benefit of eliminating cancer. This will happen if the negative health impact of spending less on nutrition, hygiene, better social conditions, and so forth outweighs any positive impact such as a reduction in the use of cars.
Even a 50% discount from current prices for new cancer drugs in conjunction with the consideration of generic/biosimilar entry would still imply that 27% of income in Germany is spent on health care. To reduce this share to let’s say 20% of income it would be necessary to command an 83% discount from current prices. This implies not only to bend the ‘price curve’ but a much more drastic reversal of the current trend of increasing drug prices. Hence, it may be fair to say that, taken to an extreme, the R&D cost argument as the fundamental justification for today’s prices does not align well with the presumed willingness to pay of the German SHI. It seems at least questionable that insurees would be willing to pay this amount for a cancer cure in order to account for R&D costs, once the portion of their income spent on health care has reached a certain threshold and significantly cuts into their non-health spending. But if the extrapolated price for a cure lacks justification as implied in this study, then it appears that current prices even for small steps towards the cure (the small gains in life expectancy) need reconsideration as well. One may invoke the notions of diminishing marginal benefit of additional life years and diminishing severity of cancer here, on the basis of which the willingness to pay for more distant steps towards the cure would be lower than for the initial steps. This stands in contrast to what is implied by the concept of diminishing marginal benefit of R&D, however, which is that later market entrants are justified in commanding higher prices. The latter principle thus suggests that current prices cannot be easily compensated by sufficiently large discounts for products entering the market later.
One may counterargue that such discounts may even be possible when envisioning a single-step cure because R&D costs of such a drug would be distributed over a large patient population. In fact, in a similar gedankenexperiment to this one, Bhattacharya et al.  assumed a single-step cure at a cost of just $10,000 per cancer patient, which was deliberately chosen to be optimistically low even at the time of their publication. But a single cure would, of course, deviate from the past history of small incremental gains in life expectancy, which the present study uses as a basis for its calculation in order to test the plausibility of the R&D cost argument. Hence, such a miracle drug seems unrealistic, at least when it comes to curing cancer as such (acknowledging that for specific types of cancer or patient subgroups a cure may be both conceivable and affordable). One may counterargue that obtaining any immediate cure – be it through a miracle drug or the sum of small incremental innovations – is unlikely and purely hypothetical. Therefore, the gedankenexperiment would fail. Yet, similar hypothetical scenarios and thought experiments are common in the health economics literature. Consider, e.g., the question posed by the time trade-off (TTO) questionnaire, which elicits quality-of-life weights and underlies one of the most common health-related quality-of-life questionnaires used in clinical research, the EuroQol five dimensions questionnaire (EQ-5D): The TTO questionnaire asks for the number of remaining life years one is willing to give up in order to be cured from, say, cancer. This is very similar to the trade-off raised by this article, viz., how much non-health consumption (in monetary terms) we as a society are willing to give up in order to be cured from cancer. That is, in both cases we capture a trade-off involving a hypothetical cure for cancer.
In addition, one may criticize the logic of taking high prices for small incremental innovations to an extreme. But again, such linear extrapolation is common in the health economics literature. For example, when eliciting the willingness to pay for a QALY in the general public by a survey, the estimate is obtained only for a fraction of a QALY in order to avoid hitting an income constraint [33,34]. The willingness-to-pay value is then extrapolated to match a full QALY [33,34]. Similarly, the calculation of the ICER extrapolates the cost of gaining less than one QALY to a full QALY using linear extrapolation. The only difference is that this study extrapolates to more than one unit of health outcome whereas the former approaches extrapolate to exactly one unit of health outcome.
Furthermore, one may counter that a growing economy would be able to accommodate future cancer drug expenditure increases. However, while price levels may be sustainable, their justification based on R&D costs fails as shown in the extreme-case scenario envisioned in this study. Hence, there is a difference between what we are able to pay on the one hand and what we are willing to pay considering the opportunity costs on the other hand.
As a word of caution, modeling studies such as this one are rarely perfect due to constraints of resources, time, and information availability. On the one hand, our model even underestimates the costs of a cancer cure because costs of drug-related AEs and drug-related services are ignored and costs of cancer treatment are limited to a period of 1 year. That is, I do not account for the fact that some cancers have a chronic course, thus mandating treatment for more than 1 year. Also, costs of curing cancer do not include a potential premium for eliminating anxiety associated with cancer (cf. ). Fully accounting for these aspects would increase the costs of a cancer cure and support the conclusions of this paper. Furthermore, cancer survivors are at increased risk for cardiovascular disease . It remains to be investigated, however, whether an increase in expenditure for cardiovascular disease (and thus the costs of a cancer cure) is offset by less spending on non-cardiovascular disease due to earlier death. On the other hand, costs of a cancer cure are overestimated because separate modelling of expenditure data for survivors and decedents as opposed to using age-specific average cost data would decrease life extension costs associated with the elimination of cancer mortality cf. . Also, the survival benefit is underestimated as it is confined to the trial period . Therefore, the current exchange rate between money and health is overestimated and so is the cost of a cancer cure. Some of the biases mentioned in this and the previous paragraph may cancel out, however.
Arguably, a more comprehensive assessment of the health gain from cancer elimination could be obtained through the QALY metric, which combines survival with a valuation of health-related quality of life . I did not calculate QALYs, however, due to a lack of aggregated data on cancer-related quality of life. If currently available treatments reduced the morbidity and mortality burden of cancer to the same degree, results would be exactly the same as for the calculation of life years (because each life year gained would be associated with a proportional quality-of-life improvement both for the current and the remaining burden reduction). Yet, if the impact of current cancer drugs were smaller on the morbidity burden, a cancer cure would result in even higher expenditure when using the QALY metric. The reason is that the remaining morbidity burden that would need to be eliminated by a cure would become larger, resulting in more QALYs gained by a cure and higher expenditures based on the fixed exchange rate between money and health. In any case, even the QALY metric is not able to fully capture an elimination of cancer-associated anxiety.
Transferability of the results from a German setting to other jurisdictions depends, among others, on how the population burden of cancer, costs of new cancer drugs, and health care expenditures as a percentage of income compare to Germany. Taking the U.S. as an example, spending on cancer drugs as a percentage of total drug expenditures is lower than in Germany (11.5% vs. 15.9%)  but health care expenditures as a percentage of Gross Domestic Product (17.2% vs, 11.3%) and incidence of malignant neoplasms (318 vs. 284 per 100,000) are higher . Therefore, given that these differences cancel out to some degree, the results of this study may also apply to other jurisdictions such as the U.S.