Analysis of 2016 Connecticut ACO Medicare Shared Savings Program Data to Identify Opportunities for Population Health Pharmacist Services

November 30, 2018
Kathryn Steckowych, PharmD

,
Marie Smith, PharmD

,
Yan Zhuang, PhD

The American Journal of Accountable Care, December 2018, Volume 6, Issue 4

Publicly reported Medicare Shared Savings Program accountable care organization (ACO) data can be analyzed to identify cost and medication-related quality performance improvement opportunities to support pharmacist integration into ACO population health services.

ABSTRACT

Objectives: To discuss the use of publicly reported Medicare Shared Savings Program (MSSP) accountable care organization (ACO) data to determine cost and medication-related quality performance improvement opportunities, analyze Connecticut MSSP ACO cost and quality performance data to identify significant findings and directional trends, and demonstrate how clinical pharmacy leaders can use this analytical approach to facilitate conversations with ACO executive leaders for pharmacist service integration.

Methods: The 2016 MSSP ACO data set was used to analyze cost and medication-related quality performance metrics for 13 Connecticut ACOs. Research questions were formulated to identify significant (1) correlations between 30-day all-cause hospital readmission (HR) or emergency department (ED) utilization rates and ACO cost or medication-related quality metric performance and (2) associations between ACO governance and ACO cost or medication-related quality metric performance. Descriptive trends analyses were performed to explain differences in cost and medication-related quality performance according to ACO governance.

Results: Statistically significant correlations were identified between (1) HR and/or ED utilization rates and several per capita expenditures or medication-related quality performance metrics and (2) ACO governance and per capita total expenditures or per capita skilled nursing facility expenditures. Looking at directional trends, health system—governed ACOs had greater total and per capita expenditures for all cost performance metrics, except per capita physician/supplier expenditures. Physician-governed ACOs performed better on all medication-related quality metrics.

Conclusions: Analysis of publicly reported MSSP ACO data can identify cost and medication-related quality performance improvement opportunities. Pharmacists can use such data for conversations with C-suite executive leaders about integrating population health pharmacist services in ACOs.

The American Journal of Accountable Care. 2018;6(4):e1-e10 In 2008, the Triple Aim was developed to “improve the experience of care, improve the health of populations, and reduce per capita costs of health care.”1 The Triple Aim guided the Center for Medicare and Medicaid Innovation to develop the first accountable care organization (ACO) model in 2012, known as the Pioneer ACO Model.2 This care delivery and healthcare payment transformation initiative allowed high-functioning healthcare organizations to test new payment models based on shared savings and population-based payments.2 As of 2016, more than 400 ACOs in the Medicare Shared Savings Program (MSSP) have been implemented.3

The MSSP is a value-based healthcare payment model that rewards healthcare organizations for delivering high-quality care while reducing total healthcare costs.4 Organizations involved in an MSSP contract must meet quality performance scores for several metrics in order to share in the generated savings with CMS. In the 2016 reporting year, MSSP ACOs were required to report year-end quality performance for 33 metrics spanning a variety of clinical foci, including medication-related process and outcome metrics.5 Many ACOs have not yet integrated pharmacists due to challenges with reimbursement for pharmacist services and identifying high-impact quality improvement opportunities for pharmacists.6

In this paper, we describe the use of publicly available MSSP ACO data in Connecticut to determine opportunities for improvement in cost and medication-related quality performance. Then, we discuss the significant findings and directional trend analyses for ACO cost and medication-related quality performance outcomes. We also describe one approach to identifying the impact of pharmacists on cost and medication-related quality by demonstrating how clinical pharmacist leaders can facilitate conversations with ACO executive leaders to explore pharmacist service integration.

METHODS

Each year, CMS publishes a public compilation of cost and quality performance data for MSSP ACOs across the country.7 Data for the 2016 reporting year5 were analyzed to evaluate cost and medication-related quality performance for 13 Connecticut ACOs. Here, we explain our ACO data analysis approach.

Extraction of Connecticut ACO Data

We extracted the MSSP data5 for ACOs with networks based solely in Connecticut or ACOs with multistate networks that included Connecticut.

Review of MSSP Data Set

To ensure accurate interpretations of the data used in the analyses, we reviewed the definitions of all cost and quality performance metrics reported within the 2016 MSSP ACO data dictionary.8 This tool provides detailed explanations of each cost and quality metric, including the numerators and denominators used to report the data.

Identification of Research Questions

The following research questions guided the MSSP ACO data analyses: (1) What is the correlation between ACO cost performance and 30-day all-cause hospital readmissions (HRs) or emergency department (ED) utilization? (2) What is the correlation between ACO medication-related quality metric performance and HR rates or ED utilization? (3) What is the association between ACO cost performance and ACO governance? and (4) What is the association between ACO medication-related quality metric performance and ACO governance?

Hospital and ED utilization and cost/quality performance. Two of the aforementioned research questions explored correlations between HRs or ED utilization and (1) ACO cost performance or (2) ACO medication-related quality performance. We selected this approach because hospitalizations, including readmissions, and ED visits are large drivers of total healthcare costs.9 Additionally, 38% of the Connecticut ACOs are governed by health systems that are invested in efforts to reduce preventable HRs and unnecessary ED visits. Therefore, the impact of integrated population health pharmacists on reducing medication-related hospital and ED utilization could potentially lead to significant reductions in ACO total healthcare expenditures.

ACO governance and cost/quality performance. The remaining 2 research questions explored associations between ACO governance models and cost or medication-related quality performance metrics. This was done to determine if quality improvement opportunities varied across ACOs with different governance structures. We categorized each Connecticut ACO as either physician-governed (PG) or health system—governed (HSG) based on publicly reported governance information found on each health system’s website. PG ACOs “may include single or multiple owners (ie, physician groups/practices) but are characterized by providing outpatient care directly to their patient population and contracting with other providers to provide hospital or other services.”10 HSG ACOs “may have a single or multiple owners (ie, hospital) with at least one owner providing direct inpatient services. Outpatient services may be provided directly if the owner(s) is an integrated health system or physician-hospital organization.”10

Selection of Cost and Medication-Related Quality Metrics

The definitions of cost and medication-related quality performance metrics were reviewed to select metrics that had strong relevance to medication therapy or the potential for clinical impact by a population health pharmacist. Eight cost performance metrics and 10 medication-related quality performance metrics were used in the statistical analyses.

Data Analysis

Analysis of hospital and ED utilization and cost/quality performance. Correlation coefficient analyses were performed to determine the correlations between (1) HR rates and ACO cost or medication-related quality performance and (2) ED utilization rates and ACO cost or medication-related quality performance. Correlation coefficients (r) greater than 0.5 or less than —0.5 were further investigated for statistical significance using linear regression analyses. T tests were used to examine statistical significance for the influence of ACO cost or medication-related quality performance. P <.05 was considered statistically significant.

Analysis of ACO governance and cost/quality performance. Linear regression analysis was used to examine the influence of ACO governance on the individual cost and quality performance variables. T tests were used to examine statistical significance for the influence of ACO governance on the individual cost and quality performance variables. P <.05 was considered statistically significant.

RESULTS

Thirty-six correlation coefficient analyses were performed between (1) HR rates and cost or medication-related quality performance and (2) ED utilization rates and cost or medication-related quality performance. Thirteen correlations were statistically significant and are reported in Table 1 [part A and part B].5

Eighteen linear regression analyses were performed between ACO governance and cost or medication-related quality performance. Two statistically significant associations were found between ACO governance and cost performance and are reported in Table 15 under “Analysis of ACO Governance and Cost/Quality Performance.” No statistically significant associations were noted between ACO governance and any of the medication-related quality performance metrics.

Given the limited number of statistically significant findings for the linear regression analyses between ACO governance and cost or medication-related quality performance, we conducted a descriptive trends analysis to identify potential differences in cost performance (Table 25) and quality performance (Table 35) according to ACO governance. Overall, HSG ACOs had greater expenditures (total and per capita) for all cost performance metrics, except for per capita physician/supplier expenditures. PG ACOs performed better on all medication-related quality metrics.

DISCUSSION

The statistically significant findings of the correlation coefficient and linear regression analyses and results of the cost and medication-related quality performance descriptive trend analyses are described as follows.

Analysis of Hospital and ED Utilization and Cost Performance (Table 15)

Per capita total expenditures. Significant positive correlations were noted between (1) HRs and per capita total expenditures (r = 0.702; P = .007) and (2) ED utilization and per capita total expenditures (r = 0.708; P = .007). These results are expected given that HRs and ED visits are cost drivers for per capita total expenses.

Per capita total inpatient expenditures. A significant positive correlation was found between HRs and per capita total inpatient expenditures (r = 0.870; P <.001). As the number of HRs per patient increases, the per capita total inpatient expenditures should increase, as hospitalization expenditures are a major component of per capita total inpatient expenditures. Additionally, a significant positive correlation was noted between ED utilization and per capita total inpatient expenditures (r = 0.719; P = .042). Patients residing within the community are usually admitted to the hospital through the ED, which would increase the per capita total inpatient costs.

Per capita skilled nursing facility (SNF) expenditures. Significant positive correlations were noted between (1) HR and per capita SNF expenditures (r = 0.779; P = .002) and (2) ED utilization and per capita SNF expenditures (r = 0.755; P = .003). Both correlations are logical because patients who reside in SNFs usually have high acuity11 and are likely to experience an HR and/or ED visit to address acute care needs.

Per capita home health expenditures. A significant positive cor&shy;relation was identified between ED utilization and home health&shy;care expenditures (r = 0.666; P = .013). Home healthcare services are used for patients with complex medical needs or functional limitations to prevent use of more costly services,12 such as SNF stays or HRs. Therefore, patients who receive home healthcare are often referred to the ED for evaluation of acute care problems.

Analysis of Hospital and ED Utilization and Quality Performance Measures (Table 15)

ACO-15: pneumococcal vaccination. A significant negative correlation was noted between HRs and ACO-15 (r = —0.573; P = .041). This suggests that pneumococcal vaccinations in older adults prevent or lessen infection severity, thereby decreasing the likelihood of HRs. This may also explain the moderate negative correlation between ED utilization and ACO-15, although it was statistically insignificant (r = 0.384).

ACO-17: tobacco use/cessation. Significant negative correlations were identified between (1) HRs and ACO-17 (r = —0.725; P = .005) and (2) ED utilization and ACO-17 (r = —0.611; P = .026). This suggests that smoking cessation may prevent HRs or ED visits due to poor lung function (eg, chronic obstructive pulmonary disease, emphysema) or pulmonary infection (eg, pneumonia).

ACO-27: percentage of beneficiaries with glycated hemoglobin greater than 9%. Significant positive correlations were identified between (1) HRs and ACO-27 (r = 0.772; P = .002) and (2) ED utilization and ACO-27 (r = 0.582; P = .037). These findings imply that patients with uncontrolled diabetes are likely to be readmitted to the hospital or use the ED as a result of acute medical needs precipitated by hyperglycemia.

ACO-33: use of an angiotensin-converting enzyme inhibitor/angiotensin receptor blocker. A significant negative correlation was found between HRs and ACO-33 (r = 0.650; P = .016). This finding suggests that evidence-based prescribing of an angiotensin-converting enzyme inhibitor or angiotensin receptor blocker in patients with congestive heart failure, renal impairment, or diabetes may prevent medical condition decompensation that may otherwise precipitate an HR.

Analysis of ACO Governance and Cost/Quality Performance (Table 15)

Cost performance: per capita total expenditures. A significant association was found between ACO governance and per capita total expenditures (P = .035). HSG ACOs had a higher per capita total expenditure (median = $11,290) compared with PG ACOs (median = $10,075). Importantly, no significant differences in total attributed patient lives were noted between HSG ACOs and PG ACOs (median = 20,840 and 20,845, respectively). This implies that the number of covered lives alone is not sufficient to explain differences in per capita total expenditures based on ACO governance. These findings also suggest that HSG ACOs may have larger or broader medical specialist and provider referral networks than PG ACOs. Therefore, PG ACOs may have greater control over the use of medical resources (eg, referral to medical specialists, ED or hospital utilization) within their practice network for attributed patients.

To further assess the differences in per capita total expenditures for PG ACOs and HSG ACOs, we analyzed associations between ACO governance and several individual components of the per capita expenditure.5,8 This analysis demonstrated that HSG ACOs had greater SNF expenditures (median = $1162) compared with PG ACOs (median = $800) (P = .021). This suggests that the difference in per capita total expenditures is due to the greater per capita SNF expenditures for HSG ACOs compared with PG ACOs. This variation in SNF expenditures for HSG and PG ACOs may be due to differences in SNF referral rates or patterns during care transitions and SNF preferred networks based on cost/

quality performance.

Quality performance. No statistically significant associations were identified between ACO governance and medication-related quality performance, which suggests that ACO governance structure alone may be insufficient to identify these differences. Also, many of the medication-related quality performance metrics included in the 2016 ACO data set13 are process-oriented and are not related to patient clinical outcomes. A descriptive trend analysis of ACO governance differences in cost and quality performance is described below.

ACO Governance Descriptive Trend Analyses

ACO cost performance is shown in Table 25 and ACO quality performance is shown in Table 3.5

ACO cost performance: annualized total ACO expenditures. HSG ACOs generated greater annualized total healthcare expenditures (median = $228,374,218) compared with PG ACOs (median = $201,527,279). This is supported by the finding that HSG ACOs generated greater per capita expenditures for all individual cost expenditure variables listed in Table 25 except for “per capita physician/supplier expenditures.” Based on this finding, we examined the individual components of total expenditures to identify more specific differences between ACO governance models.

ACO cost performance: total healthcare expenditure benchmark minus total ACO expenditures. Although HSG ACOs generated greater annualized total healthcare spending, they were closer to meeting their 2016 total healthcare expenditure benchmark compared with PG ACOs. HSG and PG ACOs surpassed their annualized total healthcare spending benchmark by medians of $857,072 and $1,113,493, respectively. This trend could be corroborated by differences in cost containment opportunities between health systems and affiliated physician medical groups. Health systems that have shared governance across inpatient and outpatient services have the capability to contain both inpatient and outpatient service costs within their ACO. Conversely, PG ACOs not associated with a health system have direct control over the costs for patient care services provided only within their practice network.

ACO cost performance: achievement of Medicare shared savings. One PG ACO (ie, ACO no. 1, a single-state network in Connecticut) achieved shared savings in 2016. Notably, this ACO generated greater annualized total ACO expenditures for the 2016 calendar year ($292,645,206) compared with the Connecticut median for PG ACOs ($201,527,279). However, this ACO also generated lower per capita expenditures for most individual cost components, except for per capita home healthcare expenditures, compared with Connecticut median values for PG ACOs. It is possible that ACO no. 1 invested more per beneficiary in home healthcare services ($706) compared with other PG ACOs in Connecticut (median = $642) to avoid preventable acute care utilization, which would thereby lower the per capita total expenditure. It is also important to note that this PG ACO partnered with an integrated medical management company designed to provide clinical, administrative, and technical support for physician-led ACOs in late 2015. It is possible that the corporate oversight of a medical management company helped ACO no. 1 reduce its per capita expenditures and increase its opportunity to achieve shared savings,14 especially given that this was the first year the ACO achieved shared savings.

One of the HSG ACOs (ie, ACO no. 13, a multistate network spanning Rhode Island, Massachusetts, and Connecticut) achieved shared savings in 2016. ACO no. 13 generated one of the lowest annualized total healthcare expenditures in 2016 ($183,633,439) compared with the other HSG ACOs in Connecticut (median = $228,374,218). Additionally, this ACO generated below-median per capita expenditures for all individual cost performance variables listed in Table 2.5 It is possible that the smaller amount of attributed lives within this ACO (compared with other HSG ACOs) contributed to this ACO’s ability to manage expenditures and achieve shared savings. Additionally, ACO no. 13’s network is highly integrated, consisting of community hospitals, primary care and specialty medical groups, and a visiting nurse association (VNA). ACOs that have shared governance across outpatient medical practices, hospital systems, and community-based organizations, such as VNAs, may be in a better position to achieve shared savings with an improved ability to manage healthcare costs at all care levels for beneficiaries. However, there is inconclusive evidence in the literature regarding factors that influence ACO performance.15 We also do not know how this ACO’s costs for Connecticut providers compared with those for providers in the 2 other states.

ACO quality performance. PG ACOs performed better on all medication-related quality metrics listed in Table 35 compared with HSG ACOs. This could be supported by the finding that PG ACOs have a larger number of primary care providers (PCPs) compared with specialist providers within their ACO network. PG ACOs had a greater PCP-to-specialist ratio (median = 0.78) compared with HSG ACOs (median = 0.56). ACOs with a larger number of PCPs may be more inclined to provide whole-person chronic condition and preventive care. Therefore, ACOs that have a greater PCP-to-specialist ratio may be more likely to achieve higher overall quality performance scores for ambulatory care—sensitive chronic conditions and preventive care metrics, including those pertaining to medication use.

Practice Implications

Identifying opportunities for medication-related quality improvement. In Table 4,5 we demonstrate how a descriptive trends analysis of ACO medication-related quality performance measures can be used to identify opportunities for improvement in this area. We selected 1 PG ACO (ACO no. 5) and 1 HSG ACO (ACO no. 9) to illustrate variances in ACO-specific medication-related performance measures compared with the median value for each measure. For both ACOs, the top 3 variances in medication-related performance measures show areas with the greatest opportunity for improvement. This approach can help pharmacist leaders align ACO pharmacist service integration opportunities with the greatest needs for medication-related performance improvement for a particular ACO in their state.

Facilitating discussions with C-suite executives. The approach shown in Table 45 can generate data to engage C-suite executive leaders in discussions about integrating pharmacist services in ACOs. Pharmacy leaders can use this approach to propose strategies for pharmacist service integration to support their role in population health management to prevent readmissions and inappropriate use of medications and to optimize medication use and safety for chronic diseases. In Table 5, we exemplify this approach using the performance improvement needs identified in Table 45 for ACO no. 5 and ACO no. 9 (defined as the top 3 quality metrics with greatest magnitude for performance improvement).

Limitations

There are some limitations to the data analyses. First, the data set used to perform the statistical analyses is small and, therefore, it is possible that it fails to reveal significant correlations due to the limited number of ACOs. Second, the analysis of cost and medication-related quality performance was conducted for ACOs in a single state. Therefore, individuals looking to use this approach will need to consider that there may be wide variations in the results of the cost and quality analyses depending on the state or geographic location of the ACOs selected for analyses. Also, we made assumptions about the selected cost and quality measures that could be affected by pharmacists’ services or medication therapy based on clinical practice experiences. Finally, we did not know if ACOs had any existing pharmacist programs directed at the selected cost and quality measures. Given that most cost and quality measures had room for improvement, we believe that the analyses show opportunity for further pharmacist involvement.

CONCLUSIONS

As healthcare payment continues to move away from fee-for-service toward value-based payment models and shared savings, an opportunity exists to integrate population health pharmacists into ACO care teams. Analysis of publicly reported MSSP ACO data is one method that pharmacists can use to identify effective medication-related quality performance improvement opportunities for ACOs in their state. This approach can provide pharmacy executives and clinical pharmacy leaders with state-level data to prepare for discussions with ACO executive leaders about population health pharmacist integration.

Acknowledgments

The authors would like to acknowledge Andrew Stevens, PharmD candidate, for his assistance in the Connecticut ACO analysis project.Author Affiliations: Department of Pharmacy Practice, The University of Connecticut School of Pharmacy (KS, MS), Storrs, CT; Department of Mathematics and Statistics, Connecticut College (YZ), New London, CT.

Source of Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or nonprofit sectors.

Author Disclosures: The 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 (KS, MS); acquisition of data (MS); analysis and interpretation of data (KS, MS); drafting of the manuscript (KS, MS, YZ); critical revision of the manuscript for important intellectual content (KS, MS, YZ); statistical analysis (YZ); administrative, technical, or logistic support (MS); and supervision (MS).

Send Correspondence to: Marie Smith, PharmD, Department of Pharmacy Practice, The University of Connecticut School of Pharmacy, 69 N Eagleville Rd, Unit 3092, Storrs, CT 06269. Email: marie.smith@uconn.edu.REFERENCES

1. Berwick DM, Nolan TW, Whittington J. The triple aim: care, health, and cost. Health Aff (Millwood). 2008;27(3):759-769. doi: 10.1377/hlthaff.27.3.759.

2. Pioneer ACO Model. CMS website. innovation.cms.gov/initiatives/Pioneer-ACO-Model. Updated August 28, 2018. Accessed April 30, 2018.

3. 2016 Medicare Shared Savings Program Organizations. CMS website. data.cms.gov/Special-Programs-Initiatives-Medicare-Shared-Savin/2016-Medicare-Shared-Savings-Program-Organizations/5kdu-cnmy. Updated June 19, 2017. Accessed July 3, 2018.

4. Shared Savings Program. CMS website. cms.gov/Medicare/Medicare-Fee-for-Service-Payment/sharedsavingsprogram/index.html. Updated August 10, 2018. Accessed April 30, 2018.

5. 2016 Shared Savings Program (SSP) Accountable Care Organizations (ACO) PUF. CMS website. data.cms.gov/Special-Programs-Initiatives-Medicare-Shared-Savin/2016-Shared-Savings-Program-SSP-Accountable-Care-O/3jk5-q6dr. Updated August 30, 2018. Accessed May 8, 2018.

6. Bonner L. Pharmacists, underutilized in ACOs, making impact in them nationwide. Pharmacy Today. 2016;22(4):43. pharmacytoday.org/article/S1042-0991(16)00518-1/fulltext. Accessed April 30, 2018.

7. Shared Savings Program Accountable Care Organizations (ACO) public-use files. CMS website. cms.gov/Research-Statistics-Data-and-Systems/Downloadable-Public-Use-Files/SSPACO/. Updated August 30, 2018. Accessed April 23, 2018.

8. 2016 Shared Savings Program (SSP) Accountable Care Organizations (ACO) PUF: Dictionary.ACO.SSP.PUF.Y2016.xlsx. CMS website. data.cms.gov/api/views/3jk5-q6dr/files/4a65b759-83a3-444d-821b-cb03afeaa9f4?download=true&filename=Dictionary.ACO.SSP.PUF.Y2016.xlsx. Accessed May 8, 2018.

9. National health expenditures 2016 highlights. CMS website. cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/NationalHealthExpendData/downloads/highlights.pdf. Accessed June 11, 2018.

10. McIntyre J. Report of the Council on Medical Service: physician access to ACO participation. American Medical Association website. ama-assn.org/sites/default/files/media-browser/public/about-ama/councils/Council%20Reports/council-on-medical-service/a15-cms-report7.pdf. Published 2015. Accessed June 13, 2018.

11. Britton MC, Ouellet GM, Minges KE, Gawel M, Hodshon B, Chaudhry SI. Care transitions between hospitals and skilled nursing facilities: perspectives of sending and receiving providers. Jt Comm J Qual Patient Saf. 2017;43(11):565-572. doi: 10.1016/j.jcjq.2017.06.004.

12. Reckrey JM, Soriano TA, Hernandez CR, et al. The team approach to home-based primary care: restructuring care to meet patient, program, and system needs. J Am Geriatr Soc. 2015;63(2):358-364. doi: 10.1111/jgs.13196.

13. Accountable care organization 2016 program quality measure narrative specifications. CMS website. cms.gov/Medicare/Medicare-Fee-for-Service-Payment/sharedsavingsprogram/Downloads/2016-ACO-NarrativeMeasures-Specs.pdf. Published January 13, 2016. Accessed June 12, 2018.

14. Lewis VA, D’Aunno T, Murray GF, Shortell SM, Colla CH. The hidden roles that management partners play in accountable care organizations. Health Aff (Millwood). 2018;37(2):292-298. doi: 10.1377/hlthaff.2017.1025.

15. Ouayogodé MH, Colla CH, Lewis VA. Determinants of success in shared savings programs: an analysis of ACO and market characteristics. Healthc (Amst). 2017;5(1-2):53-61. doi: 10.1016/j.hjdsi.2016.08.002.