- Email: [email protected]
Accountable care organizations and the use of cancer screening
Preventive Medicine 101 (2017) 15–17
Contents lists available at ScienceDirect
Preventive Medicine journal homepage: www.elsevier.com/locate/ypmed
Short Communication
Accountable care organizations and the use of cancer screening Christian P. Meyer a,b, Anna Krasnova b, Jesse D. Sammon c, Stuart R. Lipsitz b, Joel S. Weissman b, Maxine Sun a,b, Quoc-Dien Trinh a,b,⁎ a b c
Division of Urological Surgery, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA Center for Surgery and Public Health, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA Center for Outcomes Research, Analytics and Evaluation, Vattikuti Urology Institute, Henry Ford Health System, Detroit, MI, USA
a r t i c l e
i n f o
Article history: Received 9 January 2017 Received in revised form 20 March 2017 Accepted 17 May 2017 Available online 18 May 2017 Keywords: Cancer screening Accountable care organization Medicare Prostate cancer Breast cancer
a b s t r a c t Cancer preventive services, when used appropriately, result in improved health, better quality of life and decreased costs. For these reasons, cancer preventive services represent important priorities within the Affordable Care Act (ACA). Among the many provisions to improve access to preventive services the ACA introduced Accountable Care Organizations (ACOs) as trajectory to deliver coordinated, high-quality care. In order to evaluate this benchmark, we analyzed (in 2016/Boston) screening prevalence of breast cancer, a recommended screening test according to the United States Preventive Services Task Force (USPSTF), and prostate cancer, for which screening is no longer recommended by the USPSTF, among traditional Medicare beneficiaries and those enrolled in ACOs. We used propensity-score weighting to adjust for baseline confounders. We found that the prevalence of breast cancer screening (35.0% vs. 25.2%, p b 0.001) and prostate cancer screening (54.6% vs. 41.7%, p b 0.001) is higher among ACO enrollees. Our results suggest increased utilization of cancer preventive care within ACOs, regardless of whether the test is recommended or not. Better efforts may be needed within the ACO infrastructure to encourage recommended preventive care, but also penalize unnecessary use of low value services. © 2017 Elsevier Inc. All rights reserved.
1. Introduction Accountable Care Organizations (ACOs) were established under the Affordable Care Act (ACA) as a new payment model to impose greater responsibility on stakeholders for cost control and quality improvement. Medicare ACOs share in savings if spending for an attributed population falls below specific benchmarks. Shared savings depend on an ACO's performance on quality measures, which includes the receipt of preventive services categorized as ‘A’ (‘high certainty of substantial net benefit’, service should be offered or provided) and ‘B’ (‘high certainty of moderate net benefit’, service should be offered or provided) recommendations according to the US Preventive Services Task Force (USPSTF). The introduction of ACOs as part of the ACA represents a major healthcare initiative for the Centers for Medicare & Medicaid Services (CMS). Early publications have focused on financial performance and care coordination within ACOs,(McWilliams et al., 2015; McWilliams et al., 2016; McWilliams et al., 2013) however less is known about the effect of ACOs on cancer preventive services. Based on these considerations, we sought to examine and compare the prevalence of breast cancer screening (BCa-S), and prostate cancer screening (PCa-S) between ⁎ Corresponding author at: Division of Urological Surgery, Brigham and Women's Hospital, 45 Francis St, ASB II-3, Boston, MA, 02115, USA. E-mail address: [email protected] (Q.-D. Trinh).
http://dx.doi.org/10.1016/j.ypmed.2017.05.017 0091-7435/© 2017 Elsevier Inc. All rights reserved.
ACO and traditional Medicare beneficiaries. We hypothesized that the use of BCa-S, a USPSTF grade B recommendation and quality measure benchmark in the Medicare Shared Savings Program, is higher among beneficiaries assigned to an ACO, whereas the use of PCa-S, a non-recommended screening test, would be unaffected by ACO assignment. 2. Methods 2.1. Study population We relied on a random 20% sample of Medicare beneficiaries with continuous enrollment for Part A and B who were not part of an Health Maintenance Organization (HMO) in 2012–2013 between January 1, 2013 and December 31, 2013. BCa-S among those aged b 75 years old was evaluated per guidelines recommendations. PCa-S among those aged b 75 years old was also assessed (non-recommended). Cancer screening was identified using International Classification of Disease (ICD)-9, Healthcare Common Procedure Coding System (HCPCS) and Common Procedural Terminology (CPT) codes. Beneficiaries b65 years as of January 1, 2012 were excluded. Specifically for women, given that the USPSTF recommends BCa-S biennially, we excluded all of those who had a mammogram in 2012, as it would have been appropriate (i.e. not wrong) to then forego the test for the year 2013. Finally, patients with a prostate or breast cancer diagnosis in the year 2012 were also excluded, given that we chose to assess screening mammograms
16
C.P. Meyer et al. / Preventive Medicine 101 (2017) 15–17
ACO coverage was ascertained from the quarterly assignment in the ACO beneficiary-level file. We focused on the year 2013, as it was the first and only calendar year available at the time of analysis to have 12-month data on ACO enrollment. In addition, we specifically only included those beneficiaries with a continuous ACO enrollment throughout the entire year of 2013 to avoid spillover effects.
Hispanic, other), Charlson-Deyo Comorbidity Index (CCI) (0,1, or ≥2) (Deyo et al., 1992) and geographic region (South, West, Midwest, Southeast). These analyses were replicated for both the BCa-S and PCa-S populations. All analyses were clustered by state as a means to adjust for hierarchical confounders. Clustering by state accounts for variation due to within state random factors. (Kalton, 1983) Analyses were performed using SAS, version 9.3 (SAS Institute, Cary, North Carolina), in 2016. Two-sided statistical significance was defined as a p-value b 0.05. An institutional review board waiver was obtained before the study was conducted.
2.3. Statistical analyses
3. Results
In the first part of our analyses, we compared baseline characteristics of ACO vs. non-ACO (i.e. traditional Medicare beneficiaries) individuals for the BCa-S and PCa-S populations. The Kruskal-Wallis and chi-square test were used to assess any statistical difference between the two treatment groups for continuously and categorical variables, respectively. Second, observed differences in baseline characteristics between individuals in an ACO vs. non-ACO Medicare were controlled for with inverse probability weighting (average treatment effect). Differences between ACO vs. non-ACO groups were evaluated using chi-square tests. Adjusted variables consisted of age, race (white, black, Asian,
The baseline characteristics of ACO vs. non-ACO beneficiaries in the BCa-S and PCa-S populations before and after propensity score weighting are described in Table 1. Following propensity score weighting, our final cohorts of ACO and non-ACO Medicare beneficiaries included 52,987 and 526,063 women for BCa-S, and 86,936 and 814,221 men for PCa-S, respectively. Following inverse probability weighting adjustment, all p-values were not significant at the 0.05 level, which indicated that patients between groups were subsequently comparable. The prevalence of screening in ACO vs. non-ACO Medicare was 35.0% vs. 25.2% for BCa-S, and 54.6% vs. 41.7% for PCa-S (Fig. 1, both p b 0.001).
and prostate-specific antigen (PSA), rather than oncological surveillance tests. 2.2. ACO vs. traditional Medicare
Table 1 Baseline characteristics of a random 20% sample of Medicare beneficiaries eligible for colorectal, breast and prostate cancer screening assigned to traditional Medicare vs. accountable care organizations before and after propensity score weighting. (Place of Study: Boston. Time of study: 2016). Breast cancer screening
Age Mean Median 66–70 71–75
Sex Male Female CCI 0 1 ≥2 Race White Black Asian Hispanic Other
Region South West Midwest Northeast
Prostate cancer screening
Pre-propensity score weighting
Post-propensity score weighting
Pre-propensity score weighting
Post-propensity score weighting
No ACO (%) 526,085
ACO (%) 52,975
p
No ACO (%) 526,063
ACO (%) 52,987
p
No ACO (%) 814,235
ACO (%) 87,119
p
No ACO (%) 814,221
ACO (%) 86,936
p
70.74 70.67 216,320 (41.12) 309,765 (58.88)
70.88 70.86 20,357 (38.43) 32,618 (61.57)
b0.001
70.76 70.69 215,098 (40.89) 310,966 (59.11)
70.76 70.70 21,514 (40.60) 31,473 (59.40)
0.75
70.66 70.56 344,713 (42.34) 469,522 (57.66)
70.81 70.77 34,448 (39.54) 52,671 (60.46)
b0.001
70.68 70.59 342,570 (42.07) 471,651 (57.93)
70.69 70.60 36,318 (41.78) 50,618 (58.22)
0.45
814,235
87,119
526,085
52,975
270,079 (51.34) 112,267 (21.34) 143,739 (27.32)
22,313 (42.12) 13,097 (24.72) 17,565 (33.16)
408,613 (50.18) 170,219 (20.91) 235,403 (28.91)
34,554 (39.66) 21,717 (24.93) 30,848 (35.41)
400,270 (49.16) 173,388 (21.30) 240,563 (29.55)
42,186 (48.53) 18,618 (21.42) 26,132 (30.06)
441,662 (83.95) 47,407 (9.01) 11,472 (2.18) 10,459 (1.99) 15,085 (2.87)
44,790 (84.55) 4488 (8.47) 1629 (3.08) 846 (1.6) 1222 (2.31)
700,843 (86.07) 61,253 (7.52) 12,048 (1.48) 12,282 (1.51) 27,809 (3.42)
76,750 (88.10) 5350 (6.14) 1586 (1.82) 863 (0.99) 2570 (2.95)
702,475 (86.28) 60,165 (7.39) 12,297 (1.51) 11,870 (1.46) 27,414 (3.37)
75,313 (86.63) 6415 (7.38) 1222 (1.41) 1174 (1.35) 2812 (3.23)
212,061 (40.31) 105,306 (20.02) 112,380 (21.36) 96,321 (18.31)
17,707 (33.43) 6365 (12.02) 15,072 (28.45) 13,830 (26.11)
316,421 (38.86) 168,574 (20.70) 181,214 (22.26) 147,997 (18.18)
28,623 (32.86) 9720 (11.16) 25,479 (29.25) 23,296 (26.74)
311,696 (38.28) 161,055 (19.78) 186,734 (22.93) 154,736 (19.00)
33,326 (38.33) 17,001 (19.56) 20,066 (23.08) 16,543 (19.03)
b0.001
b0.001
0.12
b0.001
265,629 (50.49) 113,881 (21.65) 146,553 (27.86)
26,699 (50.39) 11,412 (21.54) 14,876 (28.07)
441,964 (84.01) 47,151 (8.96) 11,880 (2.26) 10,269 (1.95) 14,800 (2.81)
44,639 (84.25) 4821 (9.10) 1125 (2.12) 986 (1.86) 1416 (2.67)
208,754 (39.68) 101,454 (19.29) 115,798 (22.01) 100,057 (19.02)
21,094 (39.81) 10,182 (19.22) 11,691 (22.06) 10,020 (18.91)
0.39
0.86
0.98
N0.99
b0.001
b0.001
0.03
b0.001
0.28
0.42
0.97
N0.99
C.P. Meyer et al. / Preventive Medicine 101 (2017) 15–17
Fig. 1. Prevalence of breast and prostate cancer screening between traditional Medicare and ACO Medicare beneficiaries following inverse probability (propensity) weighting. Medicare beneficiaries enrolled in ACOs had a significantly higher screening prevalence of recommended (breast cancer) and non- recommended screening (prostate cancer) (Place of Study: Boston. Time of study: 2016).
17
towards a two-sided risk shared savings and loss model as early as 2019, the pressure on ACO to not only provide high-quality care, but also avoid low-value care will be strengthened. On the other hand, cancer preventive services are important considering the ambitious targets set forth by Healthy People 2020. Specifically, the Healthy People 2020 initiative, started by the Office of Disease Prevention and Health Promotion, aims to reduce cancer mortality by increasing the use of evidence-based cancer screening among US adults, targeting a nationwide rate of 81% for BCa-S and improve the prostate cancer death rate by 10%. (Office of Disease Prevention and Health Promotion, 2017). Going forward, it will be important for the health care system and providers to reconcile both the need to effectively maximize cancer preventive services in order to save lives while controlling cost. Our study has limitations similar to previous contributions. (Cooper and Koroukian, 2004; Fenton et al., 2014) First, we were not able to fully account for the indications of these tests. Second, our goal was to assess screening per guideline recommendations within a one-year timeframe – not to assess screening behaviors over time. Third, the existence of conflicting guideline statements may differentially impact the use of screening tests. Finally, we could not adjust for unmeasured confounders; for example, ACO beneficiaries may be receiving care from institutions and providers with vastly different management paradigms than those who were enrolled in traditional Medicare.
4. Discussion Conflict of interest With the passage of the ACA, the US embarked on a national experiment with new health care delivery models, such as ACOs. The effect of ACOs on the use of cancer preventive services remains largely unknown. Hence, our objective was to compare the use of screening for the two commonest non-skin malignancies in the US, namely breast and prostate cancers. We hypothesized that women treated by a provider affiliated with an ACO would be more likely than women treated by a provider not affiliated with an ACO to undergo BCa-S, a guideline recommendation and ACO quality benchmark for those under the age of 75 years old who did not receive a screening mammogram the year before. On the other hand, we hypothesized that PCa-S utilization in men younger than 75 years old would not be different between ACO and non-ACO affiliated providers since this test is non-recommended according to the latest USPSTF recommendations. Our study found that the use of recommended BCa-S was higher among women assigned to an ACO compared to traditional Medicare. Yet, the use of PCa-S in men, a non-recommended test, was also greater among ACO beneficiaries. Despite the generally acknowledged variation in preventive care performance between providers (Flocke and Litaker, 2007; Solberg et al., 2001), it has previously been suggested that coordinated health care networks outpace traditional fee-for-service plans with regards to preventive care visits and preventive care quality (Chung et al., 2015; Pauly et al., 2014). In this context, our findings suggest that the ACO model affects preventive service utilization. However, it is unclear whether our findings reflect adherence to guideline recommendations or merely an improved access to health care and/or organizational characteristics. Future endeavors should aim to compare the same providers screening behaviors before and after being assigned to an ACO. As such, our results could not conclusively indicate that ACOs had any impact on the use of cancer preventive services. ACOs were conceptually designed to encourage high-quality care, and at the same time discourage low-value care. Given the early model of incentives without disincentives for preventive care benchmarks, it may not be all that surprising that we found increased utilization of cancer screening, regardless of its perceived value according to USPSTF. As Medicare progresses
The authors declare that there is no conflict of interest. Acknowledgements Quoc-Dien Trinh is supported by an unrestricted educational grant from the Vattikuti Urology Institute, a Clay Hamlin Young Investigator Award from the Prostate Cancer Foundation (16YOUN20) and a Genentech BioOncology Career Development Award from the Conquer Cancer Foundation of the American Society of Clinical Oncology (10202). References Chung, S., Lesser, L.I., Lauderdale, D.S., Johns, N.E., Palaniappan, L.P., Luft, H.S., 2015. Medicare annual preventive care visits: use increased among fee-for-service patients, but many do not participate. Health Aff. 34, 11–20. Cooper, G.S., Koroukian, S.M., 2004. Racial disparities in the use of and indications for colorectal procedures in Medicare beneficiaries. Cancer 100, 418–424. Deyo, R.A., Cherkin, D.C., Ciol, M.A., 1992. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J. Clin. Epidemiol. 45, 613–619. Fenton, J.J., Zhu, W., Balch, S., Smith-Bindman, R., Fishman, P., Hubbard, R.A., 2014. Distinguishing screening from diagnostic mammograms using Medicare claims data. Med. Care 52, e44–e51. Flocke, S.A., Litaker, D., 2007. Physician practice patterns and variation in the delivery of preventive services. J. Gen. Intern. Med. 22, 191–196. Kalton, G., 1983. Introduction to Survey Sampling. SAGE Publishing. McWilliams, J.M., Landon, B.E., Chernew, M.E., 2013. Changes in health care spending and quality for Medicare beneficiaries associated with a commercial ACO contract. JAMA 310, 829–836. McWilliams, J.M., Chernew, M.E., Landon, B.E., Schwartz, A.L., 2015. Performance differences in year 1 of pioneer accountable care organizations. N. Engl. J. Med. 372, 1927–1936. McWilliams, J.M., Hatfield, L.A., Chernew, M.E., Landon, B.E., Schwartz, A.L., 2016. Early performance of accountable care organizations in medicare. N. Engl. J. Med. 374, 2357–2366. Office of Disease Prevention and Health Promotion, 2017. Healthy People 2020. Pauly, M.V., Sloan, F.A., Sullivan, S.D., 2014. An economic framework for preventive care advice. Health Aff. 33, 2034–2040. Solberg, L.I., Kottke, T.E., Brekke, M.L., 2001. Variation in clinical preventive services. Eff. Clin. Pract. 4, 121–126.
Contents lists available at ScienceDirect
Preventive Medicine journal homepage: www.elsevier.com/locate/ypmed
Short Communication
Accountable care organizations and the use of cancer screening Christian P. Meyer a,b, Anna Krasnova b, Jesse D. Sammon c, Stuart R. Lipsitz b, Joel S. Weissman b, Maxine Sun a,b, Quoc-Dien Trinh a,b,⁎ a b c
Division of Urological Surgery, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA Center for Surgery and Public Health, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA Center for Outcomes Research, Analytics and Evaluation, Vattikuti Urology Institute, Henry Ford Health System, Detroit, MI, USA
a r t i c l e
i n f o
Article history: Received 9 January 2017 Received in revised form 20 March 2017 Accepted 17 May 2017 Available online 18 May 2017 Keywords: Cancer screening Accountable care organization Medicare Prostate cancer Breast cancer
a b s t r a c t Cancer preventive services, when used appropriately, result in improved health, better quality of life and decreased costs. For these reasons, cancer preventive services represent important priorities within the Affordable Care Act (ACA). Among the many provisions to improve access to preventive services the ACA introduced Accountable Care Organizations (ACOs) as trajectory to deliver coordinated, high-quality care. In order to evaluate this benchmark, we analyzed (in 2016/Boston) screening prevalence of breast cancer, a recommended screening test according to the United States Preventive Services Task Force (USPSTF), and prostate cancer, for which screening is no longer recommended by the USPSTF, among traditional Medicare beneficiaries and those enrolled in ACOs. We used propensity-score weighting to adjust for baseline confounders. We found that the prevalence of breast cancer screening (35.0% vs. 25.2%, p b 0.001) and prostate cancer screening (54.6% vs. 41.7%, p b 0.001) is higher among ACO enrollees. Our results suggest increased utilization of cancer preventive care within ACOs, regardless of whether the test is recommended or not. Better efforts may be needed within the ACO infrastructure to encourage recommended preventive care, but also penalize unnecessary use of low value services. © 2017 Elsevier Inc. All rights reserved.
1. Introduction Accountable Care Organizations (ACOs) were established under the Affordable Care Act (ACA) as a new payment model to impose greater responsibility on stakeholders for cost control and quality improvement. Medicare ACOs share in savings if spending for an attributed population falls below specific benchmarks. Shared savings depend on an ACO's performance on quality measures, which includes the receipt of preventive services categorized as ‘A’ (‘high certainty of substantial net benefit’, service should be offered or provided) and ‘B’ (‘high certainty of moderate net benefit’, service should be offered or provided) recommendations according to the US Preventive Services Task Force (USPSTF). The introduction of ACOs as part of the ACA represents a major healthcare initiative for the Centers for Medicare & Medicaid Services (CMS). Early publications have focused on financial performance and care coordination within ACOs,(McWilliams et al., 2015; McWilliams et al., 2016; McWilliams et al., 2013) however less is known about the effect of ACOs on cancer preventive services. Based on these considerations, we sought to examine and compare the prevalence of breast cancer screening (BCa-S), and prostate cancer screening (PCa-S) between ⁎ Corresponding author at: Division of Urological Surgery, Brigham and Women's Hospital, 45 Francis St, ASB II-3, Boston, MA, 02115, USA. E-mail address: [email protected] (Q.-D. Trinh).
http://dx.doi.org/10.1016/j.ypmed.2017.05.017 0091-7435/© 2017 Elsevier Inc. All rights reserved.
ACO and traditional Medicare beneficiaries. We hypothesized that the use of BCa-S, a USPSTF grade B recommendation and quality measure benchmark in the Medicare Shared Savings Program, is higher among beneficiaries assigned to an ACO, whereas the use of PCa-S, a non-recommended screening test, would be unaffected by ACO assignment. 2. Methods 2.1. Study population We relied on a random 20% sample of Medicare beneficiaries with continuous enrollment for Part A and B who were not part of an Health Maintenance Organization (HMO) in 2012–2013 between January 1, 2013 and December 31, 2013. BCa-S among those aged b 75 years old was evaluated per guidelines recommendations. PCa-S among those aged b 75 years old was also assessed (non-recommended). Cancer screening was identified using International Classification of Disease (ICD)-9, Healthcare Common Procedure Coding System (HCPCS) and Common Procedural Terminology (CPT) codes. Beneficiaries b65 years as of January 1, 2012 were excluded. Specifically for women, given that the USPSTF recommends BCa-S biennially, we excluded all of those who had a mammogram in 2012, as it would have been appropriate (i.e. not wrong) to then forego the test for the year 2013. Finally, patients with a prostate or breast cancer diagnosis in the year 2012 were also excluded, given that we chose to assess screening mammograms
16
C.P. Meyer et al. / Preventive Medicine 101 (2017) 15–17
ACO coverage was ascertained from the quarterly assignment in the ACO beneficiary-level file. We focused on the year 2013, as it was the first and only calendar year available at the time of analysis to have 12-month data on ACO enrollment. In addition, we specifically only included those beneficiaries with a continuous ACO enrollment throughout the entire year of 2013 to avoid spillover effects.
Hispanic, other), Charlson-Deyo Comorbidity Index (CCI) (0,1, or ≥2) (Deyo et al., 1992) and geographic region (South, West, Midwest, Southeast). These analyses were replicated for both the BCa-S and PCa-S populations. All analyses were clustered by state as a means to adjust for hierarchical confounders. Clustering by state accounts for variation due to within state random factors. (Kalton, 1983) Analyses were performed using SAS, version 9.3 (SAS Institute, Cary, North Carolina), in 2016. Two-sided statistical significance was defined as a p-value b 0.05. An institutional review board waiver was obtained before the study was conducted.
2.3. Statistical analyses
3. Results
In the first part of our analyses, we compared baseline characteristics of ACO vs. non-ACO (i.e. traditional Medicare beneficiaries) individuals for the BCa-S and PCa-S populations. The Kruskal-Wallis and chi-square test were used to assess any statistical difference between the two treatment groups for continuously and categorical variables, respectively. Second, observed differences in baseline characteristics between individuals in an ACO vs. non-ACO Medicare were controlled for with inverse probability weighting (average treatment effect). Differences between ACO vs. non-ACO groups were evaluated using chi-square tests. Adjusted variables consisted of age, race (white, black, Asian,
The baseline characteristics of ACO vs. non-ACO beneficiaries in the BCa-S and PCa-S populations before and after propensity score weighting are described in Table 1. Following propensity score weighting, our final cohorts of ACO and non-ACO Medicare beneficiaries included 52,987 and 526,063 women for BCa-S, and 86,936 and 814,221 men for PCa-S, respectively. Following inverse probability weighting adjustment, all p-values were not significant at the 0.05 level, which indicated that patients between groups were subsequently comparable. The prevalence of screening in ACO vs. non-ACO Medicare was 35.0% vs. 25.2% for BCa-S, and 54.6% vs. 41.7% for PCa-S (Fig. 1, both p b 0.001).
and prostate-specific antigen (PSA), rather than oncological surveillance tests. 2.2. ACO vs. traditional Medicare
Table 1 Baseline characteristics of a random 20% sample of Medicare beneficiaries eligible for colorectal, breast and prostate cancer screening assigned to traditional Medicare vs. accountable care organizations before and after propensity score weighting. (Place of Study: Boston. Time of study: 2016). Breast cancer screening
Age Mean Median 66–70 71–75
Sex Male Female CCI 0 1 ≥2 Race White Black Asian Hispanic Other
Region South West Midwest Northeast
Prostate cancer screening
Pre-propensity score weighting
Post-propensity score weighting
Pre-propensity score weighting
Post-propensity score weighting
No ACO (%) 526,085
ACO (%) 52,975
p
No ACO (%) 526,063
ACO (%) 52,987
p
No ACO (%) 814,235
ACO (%) 87,119
p
No ACO (%) 814,221
ACO (%) 86,936
p
70.74 70.67 216,320 (41.12) 309,765 (58.88)
70.88 70.86 20,357 (38.43) 32,618 (61.57)
b0.001
70.76 70.69 215,098 (40.89) 310,966 (59.11)
70.76 70.70 21,514 (40.60) 31,473 (59.40)
0.75
70.66 70.56 344,713 (42.34) 469,522 (57.66)
70.81 70.77 34,448 (39.54) 52,671 (60.46)
b0.001
70.68 70.59 342,570 (42.07) 471,651 (57.93)
70.69 70.60 36,318 (41.78) 50,618 (58.22)
0.45
814,235
87,119
526,085
52,975
270,079 (51.34) 112,267 (21.34) 143,739 (27.32)
22,313 (42.12) 13,097 (24.72) 17,565 (33.16)
408,613 (50.18) 170,219 (20.91) 235,403 (28.91)
34,554 (39.66) 21,717 (24.93) 30,848 (35.41)
400,270 (49.16) 173,388 (21.30) 240,563 (29.55)
42,186 (48.53) 18,618 (21.42) 26,132 (30.06)
441,662 (83.95) 47,407 (9.01) 11,472 (2.18) 10,459 (1.99) 15,085 (2.87)
44,790 (84.55) 4488 (8.47) 1629 (3.08) 846 (1.6) 1222 (2.31)
700,843 (86.07) 61,253 (7.52) 12,048 (1.48) 12,282 (1.51) 27,809 (3.42)
76,750 (88.10) 5350 (6.14) 1586 (1.82) 863 (0.99) 2570 (2.95)
702,475 (86.28) 60,165 (7.39) 12,297 (1.51) 11,870 (1.46) 27,414 (3.37)
75,313 (86.63) 6415 (7.38) 1222 (1.41) 1174 (1.35) 2812 (3.23)
212,061 (40.31) 105,306 (20.02) 112,380 (21.36) 96,321 (18.31)
17,707 (33.43) 6365 (12.02) 15,072 (28.45) 13,830 (26.11)
316,421 (38.86) 168,574 (20.70) 181,214 (22.26) 147,997 (18.18)
28,623 (32.86) 9720 (11.16) 25,479 (29.25) 23,296 (26.74)
311,696 (38.28) 161,055 (19.78) 186,734 (22.93) 154,736 (19.00)
33,326 (38.33) 17,001 (19.56) 20,066 (23.08) 16,543 (19.03)
b0.001
b0.001
0.12
b0.001
265,629 (50.49) 113,881 (21.65) 146,553 (27.86)
26,699 (50.39) 11,412 (21.54) 14,876 (28.07)
441,964 (84.01) 47,151 (8.96) 11,880 (2.26) 10,269 (1.95) 14,800 (2.81)
44,639 (84.25) 4821 (9.10) 1125 (2.12) 986 (1.86) 1416 (2.67)
208,754 (39.68) 101,454 (19.29) 115,798 (22.01) 100,057 (19.02)
21,094 (39.81) 10,182 (19.22) 11,691 (22.06) 10,020 (18.91)
0.39
0.86
0.98
N0.99
b0.001
b0.001
0.03
b0.001
0.28
0.42
0.97
N0.99
C.P. Meyer et al. / Preventive Medicine 101 (2017) 15–17
Fig. 1. Prevalence of breast and prostate cancer screening between traditional Medicare and ACO Medicare beneficiaries following inverse probability (propensity) weighting. Medicare beneficiaries enrolled in ACOs had a significantly higher screening prevalence of recommended (breast cancer) and non- recommended screening (prostate cancer) (Place of Study: Boston. Time of study: 2016).
17
towards a two-sided risk shared savings and loss model as early as 2019, the pressure on ACO to not only provide high-quality care, but also avoid low-value care will be strengthened. On the other hand, cancer preventive services are important considering the ambitious targets set forth by Healthy People 2020. Specifically, the Healthy People 2020 initiative, started by the Office of Disease Prevention and Health Promotion, aims to reduce cancer mortality by increasing the use of evidence-based cancer screening among US adults, targeting a nationwide rate of 81% for BCa-S and improve the prostate cancer death rate by 10%. (Office of Disease Prevention and Health Promotion, 2017). Going forward, it will be important for the health care system and providers to reconcile both the need to effectively maximize cancer preventive services in order to save lives while controlling cost. Our study has limitations similar to previous contributions. (Cooper and Koroukian, 2004; Fenton et al., 2014) First, we were not able to fully account for the indications of these tests. Second, our goal was to assess screening per guideline recommendations within a one-year timeframe – not to assess screening behaviors over time. Third, the existence of conflicting guideline statements may differentially impact the use of screening tests. Finally, we could not adjust for unmeasured confounders; for example, ACO beneficiaries may be receiving care from institutions and providers with vastly different management paradigms than those who were enrolled in traditional Medicare.
4. Discussion Conflict of interest With the passage of the ACA, the US embarked on a national experiment with new health care delivery models, such as ACOs. The effect of ACOs on the use of cancer preventive services remains largely unknown. Hence, our objective was to compare the use of screening for the two commonest non-skin malignancies in the US, namely breast and prostate cancers. We hypothesized that women treated by a provider affiliated with an ACO would be more likely than women treated by a provider not affiliated with an ACO to undergo BCa-S, a guideline recommendation and ACO quality benchmark for those under the age of 75 years old who did not receive a screening mammogram the year before. On the other hand, we hypothesized that PCa-S utilization in men younger than 75 years old would not be different between ACO and non-ACO affiliated providers since this test is non-recommended according to the latest USPSTF recommendations. Our study found that the use of recommended BCa-S was higher among women assigned to an ACO compared to traditional Medicare. Yet, the use of PCa-S in men, a non-recommended test, was also greater among ACO beneficiaries. Despite the generally acknowledged variation in preventive care performance between providers (Flocke and Litaker, 2007; Solberg et al., 2001), it has previously been suggested that coordinated health care networks outpace traditional fee-for-service plans with regards to preventive care visits and preventive care quality (Chung et al., 2015; Pauly et al., 2014). In this context, our findings suggest that the ACO model affects preventive service utilization. However, it is unclear whether our findings reflect adherence to guideline recommendations or merely an improved access to health care and/or organizational characteristics. Future endeavors should aim to compare the same providers screening behaviors before and after being assigned to an ACO. As such, our results could not conclusively indicate that ACOs had any impact on the use of cancer preventive services. ACOs were conceptually designed to encourage high-quality care, and at the same time discourage low-value care. Given the early model of incentives without disincentives for preventive care benchmarks, it may not be all that surprising that we found increased utilization of cancer screening, regardless of its perceived value according to USPSTF. As Medicare progresses
The authors declare that there is no conflict of interest. Acknowledgements Quoc-Dien Trinh is supported by an unrestricted educational grant from the Vattikuti Urology Institute, a Clay Hamlin Young Investigator Award from the Prostate Cancer Foundation (16YOUN20) and a Genentech BioOncology Career Development Award from the Conquer Cancer Foundation of the American Society of Clinical Oncology (10202). References Chung, S., Lesser, L.I., Lauderdale, D.S., Johns, N.E., Palaniappan, L.P., Luft, H.S., 2015. Medicare annual preventive care visits: use increased among fee-for-service patients, but many do not participate. Health Aff. 34, 11–20. Cooper, G.S., Koroukian, S.M., 2004. Racial disparities in the use of and indications for colorectal procedures in Medicare beneficiaries. Cancer 100, 418–424. Deyo, R.A., Cherkin, D.C., Ciol, M.A., 1992. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J. Clin. Epidemiol. 45, 613–619. Fenton, J.J., Zhu, W., Balch, S., Smith-Bindman, R., Fishman, P., Hubbard, R.A., 2014. Distinguishing screening from diagnostic mammograms using Medicare claims data. Med. Care 52, e44–e51. Flocke, S.A., Litaker, D., 2007. Physician practice patterns and variation in the delivery of preventive services. J. Gen. Intern. Med. 22, 191–196. Kalton, G., 1983. Introduction to Survey Sampling. SAGE Publishing. McWilliams, J.M., Landon, B.E., Chernew, M.E., 2013. Changes in health care spending and quality for Medicare beneficiaries associated with a commercial ACO contract. JAMA 310, 829–836. McWilliams, J.M., Chernew, M.E., Landon, B.E., Schwartz, A.L., 2015. Performance differences in year 1 of pioneer accountable care organizations. N. Engl. J. Med. 372, 1927–1936. McWilliams, J.M., Hatfield, L.A., Chernew, M.E., Landon, B.E., Schwartz, A.L., 2016. Early performance of accountable care organizations in medicare. N. Engl. J. Med. 374, 2357–2366. Office of Disease Prevention and Health Promotion, 2017. Healthy People 2020. Pauly, M.V., Sloan, F.A., Sullivan, S.D., 2014. An economic framework for preventive care advice. Health Aff. 33, 2034–2040. Solberg, L.I., Kottke, T.E., Brekke, M.L., 2001. Variation in clinical preventive services. Eff. Clin. Pract. 4, 121–126.