The high price of cancer treatment often grabs headlines. But how much have patients benefited from these new treatments. A paper by Howard et al. (2016) look at new cancer treatments for chronic myeloid leukemia (CML), metastatic kidney, breast, or lung tumors and they generally find that the answer is ‘yes’, recent anti-cancer treatments have delivered significant value to patients.
The authors use SEER-Medicare data looking at patients diagnosed with these cancers between 1996-2011; the authors can track survival through 2013 with these data. The benefits of the treatment are captured based on survival gains; the cost of the treatments is measured from the point of diagnosis until death using the so-called phase-of-care approach. The authors compared changes in life expectancy and lifetime medical costs for patients with breast or lung cancer who were treated with physician-administered anticancer drugs and those who were not treated with such drugs.
Using this approach, the authors found:
Breast cancer: “…life expectancy increased over time by 13.2 months, and lifetime medical costs increased by $72,200…Among those who did not receive physician-administered drugs, life expectancy increased by 2.0 months, and costs increased by $8,900. The life expectancy increase in this group could be attributable to improvements in supportive care or lead-time bias.
Lung cancer: “…life expectancy increased over time by 3.9 months and $23,200 dollars [for those receiving anti-cancer drugs]…while remaining basically unchanged for patients who did not receive such drugs.”
Kidney Cancer: “Life expectancy among patients with kidney cancer increased by 7.9 months, and lifetime costs increased by $44,700.”
CML: “Patients with CML experienced the largest gain in life expectancy (22.1 months)…lifetime medical costs…increased by $142,200, of which $126,300 was attributable to Part D spending.”
The gains in life expectancy and cost for CML were due almost entirely to the introduction of Gleevac (imatinib) in 2001.
Overall, the incremental cots per QALY generally ranged between $100,000 and $150,000; which is generally seen to be cost effective for cancer care treatments.
The study is interesting throughout. It will be interesting to re-run a similar analysis using treatments released in the last few years as well.
The authors conclusion is interesting as well:
Our results also raise the question of whether back-of-the-envelope calculations based on drugs’ prices and the survival benefits reported in clinical trials provide an accurate measure of value.
Why do real-results differ from those in clinical trials? Howard and co-authors propose four reasons
- RCTs use protools, real-world physicians do not. This could mean that the real world outcomes are better if the real-world practices are superior to those used in the initial RCT, or outcomes could be worse if the real-world physicians don’t follow best practices.
- Learning by doing. In the real world, physicians may learn through experience how to better manage side effects through altering dosing quantity or frequency. Physician experience may allow patients to remain on therapies longer than would be the case in the RCT.
- Patients in real world differ from those in an RCT. Oftentimes, RCT protocols restrict the patient population, for instance often eliminating individuals with multiple tumor sites or a number of comorbid health conditions.
- RCT adherence is typically better than real world adherence. Although adherence to cancer treatment is generally very high compared to other diseases, adherence is typically lower in the real-world in part because there are fewer physician visits as would be mandated in an RCT.
- Howard, David H., Michael E. Chernew, Tamer Abdelgawad, Gregory L. Smith, Josephine Sollano, and David C. Grabowski. “New Anticancer Drugs Associated With Large Increases In Costs And Life Expectancy.” Health Affairs 35, no. 9 (2016): 1581-1587.
Another reason, not listed here, is that the delta is taken compared to a control group that has crossover to the new therapy upon failure. Let’s say 85% of the control group doesn’t respond to their legacy, non selective therapy, and are quickly rolled into the new therapy group. This won’t effect necessarily affect PFS comparisons, but can strongly impact the scale of OS comparisons, especially if the new drug is quite effective.
Certainly relevant in some other clinical contexts is real world flexibility to drop or continue a non performing drug. This is most obvious for weight-loss drugs; if it works well in 20% of patients, it may fail an RCT but in the real world, those 20% are the ones who would stay on it. This type of effect might be contributory in SEER data too.