Mailhot Vega et al 2016
Int J Radiat Oncol Biol Phys · 2016 · United States · Proton
This cost-utility study used a Markov cohort model with three health states (healthy, alive with coronary heart disease, and dead) to compare proton with conventional photon radiotherapy for breast cancer. The model used hypothetical cohorts of women aged 40, 50 and 60, with and without cardiac risk factors, and assessed a lifetime horizon from the societal perspective in the United States. The authors concluded that proton therapy can be cost-effective in some scenarios, most plausibly for women with at least one cardiac risk factor whose photon plans cannot bring mean heart dose below 5 Gy, but not in women without cardiac risk factors at the base-case willingness-to-pay threshold.
What it compared
Proton radiotherapy versus conventional photon radiotherapy for hypothetical US women aged 40, 50 and 60, with and without cardiac risk factors, modelled over a lifetime.
What it found
The study reports no incremental cost-effectiveness ratio values. Proton therapy was judged cost-effective against willingness-to-pay thresholds of US$50,000/QALY (base case) and US$100,000/QALY, with the result depending on mean heart dose and cardiac risk factors; for women with at least one cardiac risk factor, the base-case minimum mean heart dose at US$100,000/QALY was about 6 Gy.
What to keep in mind
The conclusion is conditional on mean heart dose and cardiac risk factor status, the cohorts are hypothetical and modelled rather than real patients, the price year is not stated, the discount rate is not stated, and the study's subgroup descriptions are less specific than the abstract on the thresholds involved.
The figures it reported
This study reports no incremental cost-effectiveness ratio. That is not an omission in our table: some evaluations report dominance, a cost per life-year, or a decision-analytic result instead, and the summary above says which.
Figures are shown exactly as the paper gives them, in its own currency and price year. We do not convert between currencies: see the glossary for why.
- Patients modelled or enrolled
- Hypothetical cohorts of women aged 40, 50 and 60, with and without CRFs. No real patients; inputs from published sources (mean heart dose complication data based on Macdonald and Fagundes).
- Comparator
- Conventional photon therapy
- Indication
- Breast cancer
- Study design
- Cost-utility analysis using a Markov cohort model (health states: healthy, alive with coronary heart disease, dead)
- Time horizon
- Lifetime: from age 40, 50 or 60 until death or age 100
- Delivery technique
- not reported by this study
- Perspective
- Societal
- Currency
- US$
- Price year
- not reported by this study
- Willingness-to-pay threshold
- US$50,000/QALY (base case); US$100,000/QALY
- Discount rate
- not reported by this study
- Cost of the particle course assumed
- US$21,933 (treatment cost as recorded in the appendix)
- Cost of the comparator assumed
- US$13,552 (treatment cost as recorded in the appendix)
Whose costs were counted: a payer, a hospital, or society as a whole.
The year the costs are expressed in. Without it, a figure cannot be compared across time.
The condition attached to the finding
Women without cardiac risk factors (CRFs): proton therapy is not cost-effective in the base case (WTP US$50,000/QALY). At WTP US$100,000/QALY it is cost-effective if mean heart dose (MHD) is >10 Gy for the 40-year-old cohort or >9 Gy for the 50-year-old cohort. Women with CRFs: base case, cost-effective when MHD is >9 Gy (age 50) and >10 Gy (age 60). At WTP US$100,000/QALY, cost-effective when MHD is >6 Gy (ages 40 and 60) and >5 Gy (age 50). The abstract is less specific. It reports no cost-effectiveness in women without CRFs or with MHD <5 Gy. For women with ≥1 CRF it reports a base-case minimum MHD of about 6 Gy at US$100,000/QALY, and ≥5 Gy in the probabilistic sensitivity analysis.
Reporting quality
92 out of 100 on the Quality of Health Economic Studies instrument, a 16-item checklist covering the objective, the perspective, the sources of each estimate, the handling of uncertainty, the horizon and discounting, how costs were measured, the model's transparency, the discussion of bias, and whether funding was disclosed. It measures how well the study was reported, not whether its conclusion is correct: a well-reported study can still reach a conclusion a later trial overturns.