Utilization and Costs Among Oncologists Participating in a Private Insurance Shared Savings Model

ABSTRACT

Objectives: Participation in the Oncology Care Model has influenced care utilization and costs relative to nonparticipating practices. Less is known, however, about how care is potentially altered by participation in similar private payer–based models. Here, we take advantage of a natural experiment in which 2 large practices from among a network of oncology practices participated in a shared savings program (SSP) with a private insurer.

Study Design: Quasi-experimental (difference-in-differences) statistical analysis of oncology claims data.

Methods: We used monthly provider-level claims data from September 2014 through August 2017 for patients with breast, colon, and lung cancer from The US Oncology Network. Key outcome measures were monthly mean office visits, total costs, and buy-and-bill drug costs among patients with breast, colon, and lung cancer. We then compared the utilization and cost patterns, preintervention and post participation, among patients on this insurance at participating practices vs those of patients at nonparticipating practices.

Results: Monthly per-patient total costs in colon cancer and lung cancer were lower through the first year of participation in the program by $1391 (95% CI, –$2218 to –$563; P < .01) and $1050 (95% CI, –$1878 to –$222; P = .02), respectively. These savings increased for colon cancer but disappeared for lung cancer in the second year. The program appeared to have no significant impact on any costs for participants with breast cancer in either of the years we examined.

Conclusions: Our results suggest that private payer–based SSP models can be associated with reduced costs for colon cancer care. There is weaker evidence of effects in lung cancer and no evidence in breast cancer. Such heterogeneous effects can inform future model development.

Am J Manag Care. 2025;31(6):In Press

Takeaway Points

We examined the incentive effects of a shared savings model on treatment costs for breast, lung, and colon cancer. We found evidence showing reduced costs for colon and lung cancer.

• The incentive structure created by a shared savings model can lower treatment costs without affecting treatment quality.
• Savings are less likely to materialize for the treatment of diseases with more limited sets of alternative treatment options.

Cancer treatment costs are growing at an accelerating rate and are projected to increase 34% over the coming decade due to population changes alone.^1,2 Explanations for the acceleration include greater treatment intensity, with more patients being treated for cancer and for longer periods of time^3,4; increasing use of supportive agents and advanced imaging; and increasing prices of cancer treatments.^5,6 In an effort to slow the growth of cancer treatment costs, the US government and some private insurers have adopted value-based payment schemes, which alter the incentive structure faced by physicians when selecting and providing treatment so as to make overprovision and high-cost treatment options less attractive while also placing a greater emphasis on treatment quality.

Historically, health care in the US has used a fee-for-service (FFS) model, which reimburses providers for each service delivered to patients. Although the FFS model incentivizes the provision of treatment, it may do this excessively, and it does not incentivize physicians to seek out the most efficient treatment options vis-à-vis cost-effectiveness and quality. A value-based payment model is an alternative to the traditional FFS model, which is designed to align physician incentives with those of the health plan. A prominent example of a value-based payment scheme is the Oncology Care Model (OCM), which was launched in 2016 by the US government in partnership with more than 150 oncology practices nationwide. The OCM offered participating practices retrospective performance-based payments (PBPs) if they were able to meet OCM payment and quality goals. In addition, the OCM provided an unconditional $160 per-patient monthly subsidy to all participating practices, known as the Monthly Enhanced Oncology Services (MEOS) payment, which was intended to facilitate the transition of practices toward more efficient treatment alternatives.

The OCM was found to reduce costs, as intended, without lowering the quality of treatment.⁷ These cost savings were driven primarily by physicians switching to cheaper but similarly effective drugs, and the savings were most concentrated in colon, lung, lymphatic, and high-risk breast cancers.⁷ Unfortunately, these cost savings were insufficient to cover the cost of the program’s PBPs and MEOS payments, which caused the program to be a net loss for the government. Ultimately, the OCM ran for a $600 million loss and was discontinued in 2022.⁷

Whereas performance under the OCM has been well documented and largely settled, performance of private shared savings programs (SSPs) is less understood. At present, we are aware of 18 studies that have examined the impact of a private SSP on outcome quality, utilization, or spending. Of those that evaluated care quality outcomes associated with a private SSP, 3 showed no or mixed (ie, positive and negative) impacts,^8-10 5 showed null or positive effects,^11-15 and 2 showed positive effects.^16,17 Among those that evaluated utilization, 6 studies found mixed utilization effects^11,17-20 whereas 4 found no effects.^10,21-23 These mixed utilization effects lend themselves to mixed spending effects, with some studies finding increases,²⁴ some finding no changes or mixed changes in spending,^{8-10,12,13,20,22,23} and others measuring decreases in spending.^14-17,25 Thus, there are few studies at present and no clear consensus of the efficacy of private SSPs in achieving their stated objectives.

Here we examine an SSP devised by a private insurer and adopted by 2 oncology practices for 3 years. Unlike the OCM, this private SSP has only PBPs and no MEOS payments. For the private SSP, participating practitioners would receive an up-front episode-based payment, which was calculated by the insurer using historical average prices for similar episodes minus a small case management fee that the insurer retains, and the practitioner could keep the difference if they were able to treat the episode at lower cost while adhering to evidence-based best-practice treatment pathways agreed upon with the insurer. This ensured that practitioners adhered to high-quality treatment standards while providing them a financial incentive to pursue lower-cost treatment alternatives.

Because the OCM MEOS payments were unconditional whereas the private PBPs were granted only when savings were achieved and at a discounted value of the savings after the management fee, the private SSP may have had a lower chance of running at a loss. However, it is uncertain to what extent the MEOS payments facilitated the cost reductions attributed to the OCM. Therefore, although the private SSP we evaluate entailed less risk to the insurer, it may also have been less effective at containing costs.

METHODS

We took advantage of a natural experiment in which 2 large practices from among a network of oncology practices participated in a pilot SSP with a private insurer while the other practices in the network carried on business as usual. We compared participants with nonparticipants using a difference-in-differences (DID) approach. This methodology controls for many potential confounding factors, such as the emergence of new FDA approvals, new clinical indications, or generic approvals during the time that the policy was adopted, which can affect quality, utilization, and costs. We compared the utilization and cost patterns, pre- and post participation, among patients on this insurance at participating practices with those of patients at nonparticipating practices. This provides for a unique opportunity to contribute to the field’s understanding of the potential impacts of private payer–based SSP models in oncology.

Data and Study Population

We used monthly provider claims data from September 2014 through August 2017 for patients with breast, colon, and lung cancer from The US Oncology Network (USON). All patients in the sample were included and collapsed to the patient-by-provider-by-tumor type-by-month level. The observations are thus patient monthly means for specific tumor types at the provider level. Key outcomes measures are monthly mean per-patient number of office visits, monthly mean per-patient total costs, and monthly mean per-patient buy-and-bill drug costs separately among patients with breast, colon, and lung cancers.

Empirical Strategy

We plotted means for our 3 outcomes variables for patients managed by SSP-participating vs nonparticipating practices through time. We then conducted DID regressions for these outcomes separately for patients with breast, colon, and lung cancers. These estimates summarize the mean post– vs pre–program launch mean differences. The DID approach is a quasi-experimental method, as it attempts to mimic a randomized controlled trial by making use of a real-world emergence of treatment and control cohorts. Statistical methods are then used to measure relative changes between the cohorts.

The treatment group is defined by exposure to an intervention, which is the adoption of the SSP in this setting. The control group remains unexposed to the SSP throughout the study period. Under typical assumptions for the DID methodology (such as similar pre­intervention trends between treatment and control groups), changes in outcomes for the treatment group relative to the control group may be interpreted as plausibly causal impacts of the intervention.

We controlled for mean differences among practices, providers, month of care, patient characteristics such as age and sex, and several other characteristics fully specified in the eAppendix (available at ajmc.com). The model was estimated using ordinary least-squares regression. We assessed both the first-year and cumulative second-year program effects. As a robustness check, we also conducted these analyses for within-practice effects (as opposed to across-practice effects), comparing participating practices’ care for SSP patients vs non-SSP privately insured patients. This study was approved by the USON Institutional Review Board.

RESULTS

Mean Plots and Statistics

The mean plots suggest little to no effects on visits and potential reductions in costs and drug costs in the first year that partly reversed in the second year (Figure). The mean plots are also generally suggestive of similar preintervention trends; however, there are preintervention period–level differences in covariates and outcomes across practices (Table 1) and within practices (Table 2).

DID Estimates

Table 3 shows our DID estimates for the first year of the program for both the cross-practice and cross-payer specifications. Using the cross-practice specification, we found significant reductions in total costs for both lung and colon cancer program participants that are large relative to their means. Monthly per-patient total costs in colon cancer and lung cancer were lower through the first year of participation in the program by $1391 (95% CI, –$2218 to –$563; P < .01) and $1050 (95% CI, –$1878 to –$222; P = .02), respectively. Colon cancer drug costs also decreased by $2060 (95% CI, –$3279 to –$842; P < .01) (panel A). Using the cross-payer specification, we similarly found reductions in total costs for colon cancer in the first year: $605 for total costs (95% CI, –$1167 to –$42; P = .05) and $692 for drug costs (95% CI, –$1488 to $105; P = .06). However, the lung cancer effects did not replicate using the cross-payer specification.

Table 4 expands the sample to include the second year of the program. The first-year colon cancer reductions deepened through the second year of the program using the cross-practice specification: $1826 reductions for total costs (95% CI, –$3075 to –$576; P < .01) and $2424 for drug costs (95% CI, –$4711 to –$138; P = .04). The total cost savings found in the first-year sample for lung cancer did not continue when including 2 years of follow-up data (panel A). Using the cross-payer specification, total costs and drug costs for colon cancer similarly persisted and deepened to $1111 for total costs (95% CI, –$1680 to –$542; P = .014) and $2057 for drug costs (95% CI, –$3709 to $405; P = .03). However, in contrast, we additionally observed a reduction in office visits among patients with colon cancer through the second year of the program of 0.170 (95% CI, –0.248 to –0.091; P = .012). The savings also disappeared for lung cancer in the second year, where we observed decreases of 0.183 in monthly office visits (95% CI, –0.26 to –0.10; P < .01) and increases of $1371 in total costs (95% CI, –$273 to $3014; P = .08). Although these heterogeneous results were not present as statistically significant results in the within-practice specifications, the signs on the point estimates do align (panel B). The program also appears to have had no significant impact on any costs for participants with breast cancer in either of the years or comparison specifications that we examined (Tables 3 and 4).

DISCUSSION

There is general agreement in the extant literature that private SSPs mostly positively impact care cost-efficiency. However, there is ambiguity in the literature on how such programs impact costs and utilization. Our study aimed to help address this ambiguity by leveraging a quasi-experimental design, which controls for many potentially significant contemporaneous confounding factors that may be influencing the results of studies thus far. These factors include the emergence of new FDA approvals, clinical indications, or generic approvals during the time that the policy was adopted, which can affect quality, utilization, and costs, and are accounted for in our empirical design.

Our results show that private payer–based SSP models can be associated with efficiency gains that are consistent with the incentive structures that they create.¹⁷ These findings are also consistent with those of other programs that attempt to incentivize practitioners to adhere to evidence-based best-practice treatment pathways when delivering care, which also show cost reductions.^26,27 Our results also show that the beneficial impact of SSPs is unlikely to be uniform but rather may be contingent on factors such as the degree of clinically suitable lower-cost clinical substitutes, which can vary by tumor type.

Savings were consistently measured in colon cancer, consistently absent in breast cancer, and mixed in lung cancer (given that results do not hold across comparison specifications). These differences are likely driven by differences in therapeutic interchange sets and acceptable treatment pathways, given that drug or service prices did not change within the model. We were unable to measure drug-specific utilization, but it is possible that savings in the colon cancer setting could be driven by reductions in growth factor support (the use of which is at the discretion of the provider) or by more appropriate use of biologics, including adoption of then newly available biosimilar therapies such as bevacizumab. In contrast, the more limited set of costly discretionary therapies in lung cancer at the time (ie, EGFR therapies) along with approvals of expensive therapies such as pembrolizumab and nivolumab, whose rapid adoption may have mitigated substitution toward lower-cost options, may also explain the inconsistent results. Such heterogeneous program effects may inform future tumor-specific model development. Researchers should continue to analyze private value-based payment schemes as they evolve over time, as research shows it can take years for significant savings to emerge.^17,28

Limitations

Our study is limited in that we were able to evaluate the impact of a private-sector SSP on only 3 tumor types. In addition, we lacked a diverse selection of covariates to measure service utilization, which may have prevented us from finding meaningful and significant effects elsewhere (including at the drug and drug-class levels). Selection bias is always a concern with nonrandomized samples. The program was not offered to all practices in the physician network and was based on payer mix and payer relationship, but beyond that, we have limited insights into the underlying reasons for participation. Although similar preintervention period trends can help to contextualize comparisons between treatment and control groups, practices and payers that decide to enter into SSP contracts may be fundamentally different from those that do not enter into those contracts. We can speculate that reductions in certain drug classes may be driving our results (eg, group colony-stimulating factors in colon cancer), but we were unable to directly assess which specific drugs are actually accounting for the overall changes in drug costs. Payments for cost reductions were also made outside the claims data, so we were unfortunately unable to measure the overall impact of the program net of shared savings payments made to the practices. Finally, we were limited in the duration of the follow-up period and in measurements of care quality that can help to evaluate the duration of and trade-offs between program effects.

CONCLUSIONS

Value-based payment schemes have been rising in popularity in the US as an alternative to FFS payment in the health care industry that proponents hope can help bring greater efficiency to the sector. Here, we studied a private-sector SSP that incentivizes physicians to use more cost-effective treatment options while maintaining health outcomes by providing them performance payments from the savings achieved. We found evidence that such programs can reduce costs in the treatment of both colon and lung cancers. Whether such a program will produce savings is contingent upon the availability of more cost-effective treatment options that physicians may be indifferent to choosing in the absence of the incentive structure created by an SSP. The evidence that we present here suggests that an SSP can provide service-level cost-saving benefits in the treatment of colon cancer. Broad adoption by insurers of targeted SSPs such as this may help fight rapidly growing cancer treatment costs.

Acknowledgments

The US Oncology Health Services Research committee, led by Diana Verrilli, MS, and Beatrice Mautner, MSN, supported this research. Mittsy Mosshart, BBA; Beth Alvarez, MS; Pat Cayot; and Allen Kowalczyk, BS, prepared and extracted the data used in this study.

Author Affiliations: Tulane University (BW, AS), New Orleans, LA; US Oncology, McKesson (LW, JRH), The Woodlands, TX; Ontada, McKesson (BW, NR),
The Woodlands, TX.

Source of Funding: None.

Author Disclosures: Dr Walker has secondary employment at Ontada, which provided the data for this study, for unrelated work and has submitted grant applications and received grants for unrelated work at various organizations. Dr Wilfong has consulted for Pfizer, was employed by McKesson at the time of writing, is currently employed by Thyme Care, has received lecture fees from HMP, has attended the Patient-Centered Oncology Care^® Conference and Community Oncology Alliance Payer Exchange Summit, and owns stock in McKesson. Dr Robert is employed by and owns stock in McKesson. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (BW, LW, NR, JRH); acquisition of data (BW); analysis and interpretation of data (BW, AS, NR); drafting of the manuscript (BW, LW, AS); critical revision of the manuscript for important intellectual content (BW, LW, NR, JRH); statistical analysis (BW, AS); provision of patients or study materials (NR); administrative, technical, or logistic support (BW, JRH); and supervision (BW, LW, JRH).

Address Correspondence to: Brigham Walker, PhD, Tulane University, 1440 Canal St, Ste 1937, New Orleans, LA 70112. Email: bwalker6@tulane.edu.

REFERENCES

1. Mariotto AB, Enewold L, Zhao J, Zeruto CA, Yabroff KR. Medical care costs associated with cancer survivorship in the United States. Cancer Epidemiol Biomarkers Prev. 2020;29(7):1304-1312. doi:10.1158/1055-9965.EPI-19-1534

2. Mariotto AB, Yabroff KR, Shao Y, Feuer EJ, Brown ML. Projections of the cost of cancer care in the United States: 2010-2020. J Natl Cancer Inst. 2011;103(2):117-128. doi:10.1093/jnci/djq495

3. Bradley CJ, Yabroff KR, Warren JL, Zeruto C, Chawla N, Lamont EB. Trends in the treatment of metastatic colon and rectal cancer in elderly patients. Med Care. 2016;54(5):490-497. doi:10.1097/MLR.0000000000000510

4. Bradley CJ, Yabroff KR, Mariotto AB, Zeruto C, Tran Q, Warren JL. Antineoplastic treatment of advanced-stage non–small-cell lung cancer: treatment, survival, and spending (2000 to 2011). J Clin Oncol. 2017;35(5):529-535. doi:10.1200/JCO.2016.69.4166

5. Bach PB. Limits on Medicare’s ability to control rising spending on cancer drugs. N Engl J Med. 2009;360(6):626-633. doi:10.1056/NEJMhpr0807774

6. Myers MH, Ries LA. Cancer patient survival rates: SEER program results for 10 years of follow-up. CA Cancer J Clin. 1989;39(1):21-32. doi:10.3322/canjclin.39.1.21

7. Trombley M, McClellan S, Chami N, et al. Evaluation of the Oncology Care Model: Final Report. Center for Medicare & Medicaid Innovation. 2024. Accessed June 1, 2024. https://www.cms.gov/priorities/innovation/data-and-reports/2024/ocm-final-eval-report-2024-exec-sum

8. Stuart EA, Barry CL, Donohue JM, et al. Effects of accountable care and payment reform on substance use disorder treatment: evidence from the initial 3 years of the alternative quality contract. Addiction. 2017;112(1):124-133. doi:10.1111/add.13555

9. Salmon RB, Sanderson MI, Walters BA, Kennedy K, Flores RC, Muney AM. A collaborative accountable care model in three practices showed promising early results on costs and quality of care. Health Aff (Millwood). 2012;31(11):2379-2387. doi:10.1377/hlthaff.2012.0354

10. Donohue JM, Barry CL, Stuart EA, et al. Effects of global payment and accountable care on medication treatment for alcohol and opioid use disorders. J Addict Med. 2018;12(1):11. doi:10.1097/ADM.0000000000000368

11. Marsteller JA, Hsu YJ, Gill C, et al. Maryland multipayor patient-centered medical home program: a 4-year quasiexperimental evaluation of quality, utilization, patient satisfaction, and provider perceptions. Med Care. 2018;56(4):308-320. doi:10.1097/MLR.0000000000000881

12. Chien AT, Song Z, Chernew ME, et al. Two-year impact of the alternative quality contract on pediatric health care quality and spending. Pediatrics. 2014;133(1):96-104. doi:10.1542/peds.2012-3440

13. Song Z, Rose S, Chernew ME, Safran DG. Lower- versus higher-income populations in the alternative quality contract: improved quality and similar spending. Health Aff (Millwood). 2017;36(1):74-82. doi:10.1377/hlthaff.2016.0682

14. Song Z, Safran DG, Landon BE, et al. Health care spending and quality in year 1 of the alternative quality contract. N Engl J Med. 2011;365(10):909-918. doi:10.1056/NEJMsa1101416

15. Song Z, Rose S, Safran DG, Landon BE, Day MP, Chernew ME. Changes in health care spending and quality 4 years into global payment. N Engl J Med. 2014;371(18):1704-1714. doi:10.1056/NEJMsa1404026

16. Song Z, Safran DG, Landon BE, et al. The ‘Alternative Quality Contract,’ based on a global budget, lowered medical spending and improved quality. Health Aff (Millwood). 2012;31(8):1885-1894. doi:10.1377/hlthaff.2012.0327

17. Song Z, Ji Y, Safran DG, Chernew ME. Health care spending, utilization, and quality 8 years into global payment. N Engl J Med. 2019;381(3):252-263. doi:10.1056/NEJMsa1813621

18. Barry CL, Stuart EA, Donohue JM, et al. The early impact of the ‘Alternative Quality Contract’ on mental health service use and spending in Massachusetts. Health Aff (Millwood). 2015;34(12):2077-2085. doi:10.1377/hlthaff.2015.0685

19. Huskamp HA, Greenfield SF, Stuart EA, et al. Effects of global payment and accountable care on tobacco cessation service use: an observational study. J Gen Intern Med. 2016;31(10):1134-1140. doi:10.1007/s11606-016-3718-y

20. Joyce NR, Huskamp HA, Hadland SE, et al. The Alternative Quality Contract: impact on service use and spending for children with ADHD. Psychiatr Serv. 2017;68(12):1210-1212. doi:10.1176/appi.ps.201700143

21. Sharp AL, Song Z, Safran DG, Chernew ME, Fendrick AM. The effect of bundled payment on emergency department use: Alternative Quality Contract effects after year one. Acad Emerg Med. 2013;20(9):961-964. doi:10.1111/acem.12205

22. Afendulis CC, Hatfield LA, Landon BE, et al. Early impact of CareFirst’s patient-centered medical home with strong financial incentives. Health Aff (Millwood). 2017;36(3):468-475. doi:10.1377/hlthaff.2016.1321

23. Afendulis CC, Fendrick AM, Song Z, et al. The impact of global budgets on pharmaceutical spending and utilization: early experience from the alternative quality contract. Inquiry. 2014;51:0046958014558716. doi:10.1177/0046958014558716

24. Newhouse JP, Price M, Hsu J, Landon B, McWilliams JM. Delivery system performance as its financial risk varies. Am J Manag Care. 2019;25(12):e388-e394.

25. Milad MA, Murray RC, Navathe AS, Ryan AM. Value-based payment models in the commercial insurance sector: a systematic review. Health Aff (Millwood). 2022;41(4):540-548. doi:10.1377/hlthaff.2021.01020

26. Hoverman JR, Neubauer MA, Jameson M, et al. Three-year results of a Medicare Advantage cancer management program. J Oncol Pract. 2018;14(4):e229-e237. doi:10.1200/JOP.17.00091

27. Neubauer MA, Hoverman JR, Kolodziej M, et al. Cost effectiveness of evidence-based treatment guidelines for the treatment of non–small-cell lung cancer in the community setting. J Oncol Pract. 2010;6(1):12-18. doi:10.1200/JOP.091058

28. Song Z. Accountable care organizations: early results and future challenges. J Clin Outcomes Manag. 2014;21(8):364-371.