Aaron Mitchell: Exploring the Influence of Insurance Status on Chemotherapy Choices: Insights from Real-world Data

Aaron Mitchell, Genitourinary Oncologist at Memorial Sloan Kettering Cancer Center, shared a post on X/Twitter:

“Our latest study is out this week in The Oncologist
We looked at whether patients with commercial or Medicaid insurance are more likely to receive high-price (and hence, high-profit) chemotherapy treatments.

Oncologists’ billing for chemotherapy is proportional to chemotherapy price. The more expensive a drug, the more money we make. (sound backwards? It is!)

But HOW much more we make varies by insurer. And it varies a LOT. For public insurers (Medicare, Medicaid), it’s 4-6%. For commercial insurers? More like 100% (see recent work by Xiao, Joseph Ross, Cary Gross, Stacie Dusetzina, Michael McWilliams and co)

Hence, the incentive to use high-price drugs is much stronger when treating commercially insured vs. publicly insured patients. Do oncology providers respond to this incentive? That was the motivating question for our study.

We studied a set of “test cases,” wherein the addition of a biologic drug (the high-price drug) to a (much cheaper) combination of older cytotoxic agents was considered an option, but not mandatory, in meeting the contemporary standard-of-care.

(specifically, the National Comprehensive Cancer Network guidelines recommended both the high-price and low-price options equally during our study period)

The cases were:
1) mCRC Low price = FOLFOX or FOLFIRI
Hi price = [FOLFOX or FOLFIRI] +
[bevacizumab/cetuximab/panitumumab]

2) mNSCLC Low price = cytotoxic
Hi price = cytotoxic + bev
3) advanced head and neck
Low price = cytotoxic
Hi price = cetuximab +/- cytotoxic

We used a multi-payer database in the state of NC with linked cancer registry and claims data.
Outcome = treatment received (Low price vs. hi price)
Exposure = patient insurer (commercial vs. Medicaid)
Key covariate = treatment in an academic (NCI) center, vs. not

…with inverse PS-weighting to balance confounders, most importantly which of the three cancer types patients had.

Results: Unadjusted, 8.2% more commercial patients received hi-priced treatment than Medicaid patients (35.7% vs 27.5%) Adjusted, this shrank to 4.8%

…and was no longer statistically significant. Although, in the grand oncology tradition, please look at our subgroup analysis:

Commercial patients were only more likely to receive hi-cost treatment from non-NCI (eg., private practice) providers. In the large, academic (NCI) centers, commercial patients were not more likely to get high cost treatment. In fact, they appeared less likely.

This is exploratory, but is potentially consistent with the hypothesis that providers do respond to the financial incentive to use hi-cost treatment in the private practice setting, where compensation is more directly tied to billing.

We were limited by a single-state sample, and a relatively small cohort. Given the potential that we were underpowered, in my view this study isn’t definitive one way or the other, and it’s a question that I plan to continue pursuing.

This study holds a special place in my heart – it was the first HSR study I ever truly conceived of, over a decade ago, during the Clinical Epidemiology training course that junior residents at Duke IM Residency take (still do, I hope!).

It’s been a long road from there to designing, conducting, and publishing the study, and I’m proud to finally see it in print. This was also the last Aim of my ASCO Conquer Cancer, the ASCO Foundation YIA – huge thanks to my funders!

And to all my co-authors for helping make this study happen! Stacie Dusetzina, Alan Kinlaw, Hanna Sanoff, Jenny Lund and Sharon Peacock-Hinton.

Worth nothing, this study used the CIPHR linkage at UNC Chapel Hill, the kind of data infrastructure that is threatened with destruction by the new CMSGov policy limiting physical data access.”

Read further.

Source: Aaron Mitchell/X