Will ACOs Shift Costs to Private Insurers?

The Medicare pay-for-performance program for qualified Accountable Care Organizations (ACO) is designed to reward ACOs for decreasing Medicare fee-for-service costs of their assigned members. If the ACOs meet certain quality benchmarks, then the ACO and Medicare will share in the savings as a sort of “performance bonus.”

In an excellent New York Times article last week, reporter Robert Pear quotes an experience antitrust lawyer, J. Thomas Rosch, as stating that “there will be a real risk that the savings accruing to Medicare will just come at the expense of private insurers.”

Is this really a major risk, though? The answer is … not really, but it depends.

In an article to be published in Milbank Quarterly next month, researcher Austin Frakt, Ph.D. examines the existing literature on cost shifting.  He writes that “cost shifting can and has occurred, but usually at a relatively low rate.” He writes that “it is likely to be at a rate closer to 20 cents on the dollar than the dollar-for-dollar one suggested by industry-funded reports (PWC 2009) and by Cutler’s (1998) estimate using data from the 1985-1990 period.”

While ACO Shared Savings programs may lead to some cost shifting, the effect should not be over stated. Other factors are important in the evaluation of cost-shifting, such as the hospitals’ ability to cut costs, public/private payer mix, hospital competition, and health plan market power.

Dr. Frakt states that the debate about cost shifting amongst policymakers “provides a false impression that cost shifting is a large and pervasive phenomenon.” This conventional wisdom should be “taken with a grain of salt.” Good stuff ... check out his blog, the Incidental Economist.

The Theory of Healthcare P4P

The challenge for healthcare regulators, such as U.S. federal government’s Centers for Medicare and Medicaid Services (CMS), is to create efficient contracts, payment methodologies and incentive systems that reward improved health outcomes. Agency theory provides theoretical support for assessment of such contracts. The theory was developed separately in the early 1970s by Stephen A. Ross for economics and Barry M. Mitnick for political science and extended by Kathleen M. Eisenhardt to organizational research in the late 1980s.

 Agency theory pertains to difficult contracting or regulatory conditions where the goals between principal and agent are incongruent (Eisenhardt, 1989), the outcomes are uncertain (Ross, 1973), the agent is self-interested and risk-averse (Eisenhardt, 1989), and the principal maintains a formal oversight relationship to the agent (Mitnick, 1982).  These theoretical attributes, manifested in a negative sense as the “agency problem,” and the conditions needed to control them, such as incentives, are at the heart of this healthcare pay-for-performance contracts.

Recent healthcare research predominantly applies the theory to physicians or hospitals as agents (Cangialose, Cary, Hoffman, & Ballard, 1997; Schneider & Mathios, 2006). For care management organizations (CMOs), such as HMOs and the like, Jacob Glazer and Thomas G. McGuire (2002) argued that capitation payment made to CMOs can be manipulated “to induce the profit maximizing plan (the agent) to provide the efficient quality (the regulator’s goal).”

A regulator contracts with CMOs to obtain for improved health outcomes for members, and the CMO is expected to perform services at an agreed upon rate. But how does the regulator know that the CMOs are performing as required?

Under agency theory, an efficient contract should minimize “the principal’s monitoring and enforcement activities” (Mitnick, 1973). High cost to the regulator when conducting audits, collecting information on performance, and devising regulations and communicating performance programs is a major problem in regulatory oversight. Since it is expensive for regulators to oversee the implementation of the required programs, incentive programs can be used to try to control the behaviors of CMOs (Maio, Goldfarb, C. Carter, & Nash, 2003). A typical incentive system retrospectively pay CMOs through bonuses, fines or tiered differential payments based on population-based performance. There is little formal research on the results of these P4P schemes for CMOs.

Do CMOs affect healthcare quality independent of direct healthcare providers, such as hospitals and doctors? The preliminary answer is yes. Researchers found that increased quality of care (QOC) indicators, such as medication adherence, cancer screening, and diabetes care management, were associated with CMOs independent from the direct care given by provider groups.

An example of a successfully implemented Medicaid MCO incentive program is the New York State Quality Assurance Reporting Requirement (QARR) and Quality Incentive (QI) programs.  According to the report, evidence of QI program effectiveness includes improvement in quality of care process measure scores, the exit of poorly performing plans from the Medicaid managed care program, and improved performance scores (e.g., immunizations, well-child visits, and diabetes control) compared to Medicaid fee-for-service.

However, according to Rosenthal et al. (2005), the evidence is mixed on provider-focused P4P in terms of creating relative improvements, and the impact on their long-term effects. Rosenthal et al. (2006) argue that the small size of the bonuses account for the small effect sizes in P4P studies. Also, unintended consequences schemes may dampen the enthusiasm for P4P. For example, one study showed that the proportion of African-Americans patients treated by a hospital was inversely associated with performance for certain process-related performance measures.

Agency theory supports the use of incentive-based contracts in difficult agent-principal relations in political science, economics, and organizational research. However, the jury is still out on the effectiveness of P4P in healthcare. Certainly, the relationship of principal and agent is complicated and the oversight to ensure contract compliance is extensive. Yet, there are many extenuating issues that make the application of agency theory to healthcare P4P complicated. More investigation into the application agency theory in healthcare incentive-based contracts is required.

ACOs Impact on Nursing

As we await CMS proposed regulations on Accountable Care Organizations (ACOs), hospitals grapple with the many issues of pay-for-performance programs. I’ve posted about concerns regarding the ACO membership and quality measures. Also, I made an observation I called the ACO Paradox.

To add another issue to the list, a recent article by Kurtzman et al. (2011) in Health Affairs raises concerns about pay-for-performance (P4P) programs’ effect on the nursing workforce. The authors write that these incentive programs “increase both the burden and the blame for nurses without corresponding improvements in staffing levels, work environment, salaries, or turnover.” In addition to the Medicare hospital-acquired conditions policy and hospital inpatient value-based purchasing program, the impending ACO payment policies may add to nurse workload.

Especially for nursing-sensitive indicators (e.g., pressure ulcers, patient falls, catheter-associated infections), improved quality may be directly related to increased nurse staffing. This brings to mind two questions regarding ACOs and nurses.

First, under the health reform law, ACOs must meet certain quality thresholds to qualify for “shared savings” with Medicare. Can nursing-sensitive quality be achieved while reducing costs? Correlational research links higher nurse-to-patient staffing ratios to improved quality for some measures, including pressure ulcers, falls with injuries, catheter-associated urinary tract infections, and vascular catheter-associated infections. Since nursing can represent 40% of hospitals’ direct care budgets, achieving high marks for these types of measures may contribute to the ACO Paradox: What if improved healthcare quality is accompanied by increased costs?

Second, how should ACOs distribute potential Medicare “shared savings” bonuses among nursing staff? The Kurtzman et al. (2011) article makes the point that “performance incentives are typically paid to physicians, hospitals, and other providers, rather than directly to staff nurses—individually or collectively.” Complexity of nurse staffing at hospitals is the primary reason incentives are not paid to nurses. But should they be? Does it makes sense to reward nurses for quality?

As ACOs create legal entities to meet the CMS requirements, nurse incentives will need to be addressed.

Do the CMS hospital P4P regs signal ACO measures to come?

CMS released the proposed rules for the hospital inpatient value-based purchasing program, in support of Section 3001(a) of the Patient Protection and Affordable Care Act. The program, designed to reward hospitals for quality improvement, will apply in 2013 to payments for discharges occurring on or after Oct. 1, 2012. According to a CMS press release, this is an example of value-based purchasing (VBP) or pay-for-performance (P4P) that will move our healthcare system, “toward rewarding better value, outcomes, and innovations instead of merely volume.” CMS will accept public comments on the proposed rule through March 8th.

Do these Hospital VBP program measures provide some insight as to which measures will be selected by CMS for ACOs under the Shared Savings P4P program? If so, what can we learn?

In their comments, CMS acknowledges that these P4P systems “should rely on a mix of standards, process, outcomes, and patient experience measures, including measures of care transitions and changes in patient functional status.” This is a wise approach that I advocated in my posting, “ACO Quality: Don’t Forget the Processes.”

The 2013 Hospital VBP program will include measures already adopted for the Hospital Inpatient Quality Reporting Program (IQR).  In this Hospital IQR program, hospitals that do not participate in reporting measures receive an annual 2.0 percentage point reduction in their Medicare rate inflation adjustments (market basket update).

The majority of the Hospital VBP program proposed measures will be clinical process of care measures, such as whether aspirin was prescribed at discharge to a patient recovering from an acute myocardial infarction. Other categories for the 17 different processes of care measures include heart failure, pneumonia, healthcare-associated infections, and surgeries. In addition, a patient satisfaction survey called, the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS), will be included.  For these measures, CMS proposes that the 2013 payment be based on a three-quarter performance period from July 1, 2011 through March 31, 2012 compared to the three-quarter baseline period from July 1, 2009 to March 31, 2010.

Finally, CMS proposed to include three outcome measures in the Hospital VBP program. Instead of the three-quarter performance period proposed for the process of care and HCAHPS measures, the risk-adjusted mortality outcome measures will be aggregated for 18 months to provide sufficiently accurate information about a hospital's outcomes on which to score hospitals on these measures and base payment. CMS will use the 18-month period from July 1, 2011 to December 31, 2012 to compare to the baseline period of July 1, 2008 to December 31, 2009. Acute Myocardial Infarction 30-Day Mortality Rate (MORT-30-AMI), Heart Failure 30-Day Mortality Rate (MORT-30-HF), Pneumonia 30-Day Mortality Rate (MORT-30-PN ).

So, what’s missing? According to ACA (sec. 3001(a)), CMS could not include any measure that wasn’t already included on the Hospital Compare website for at least one year. This means that readmission measures were excluded for 2013, but CMS hopes to include those in the future. Such restrictions do not apply to the ACO Shared Savings program.

Also, CMS removed the so-called “topped-out” measures from the list. “Topped-out” measures are those where all but a few hospitals performed well and provide no meaningful differentiation between hospital quality performances. Some of these measures include aspirin at arrival (AMI-1) and beta blocker at discharge (AMI-5).

In addition, the Hospital IQR structural measures, such as participation in Stroke Registries, were excluded due to measurement and reporting problems. 

CMS did not yet propose efficiency measures, including measures of “Medicare Spending per beneficiary” or “internal hospital efficiency,” as required by statute.  CMS wants public comment as to on potential efficiency measures.

Preliminary ACO regulations are due from CMS soon. We’ll see what impact these choices have on their proposed measures.

MLR Floors Are Misguided

Enforcing the medical loss ratio (MLR) provision of the Health Reform law means that health insurers must spend 80% to 85% of premiums on medical care. In my November 2010 post regarding MLRs, I noted that preliminary CMS regulations deemed that cost control efforts would not count as “medical care” – even those activities designed to detect inappropriate services .

MLR is a basic ratio. The premium (money the health insurer is paid) is the denominator and medical care (claims, case management etc.) is the numerator. Under the law, administrative activities are not counted in the numerator. As such, efforts to reduce inappropriate childhood tube insertions or unnecessary coronary artery bypass graft (CABG) surgeries are counted against the insurer in the calculation of MLR. These activities, instead, are counted in the “other non-claims costs” category. 

From a regulator’s point of view, this makes sense. How would a state insurance department be able to sort out whether an activity was a) restricting access to appropriate care to profit or b) controlling inappropriate services by profit-seeking providers? 

Unfortunately, by classifying all direct medical services as “value to the consumer,” the law fails to recognize what the Dartmouth Atlas of Health Care research has been telling us for decades: more healthcare is not necessarily good healthcare. The failure to distinguish necessary medical care from inappropriate care makes the MLR provision wholly misguided.

The MLR floor is an understandable reaction to the abuses by some health insurers who have blatantly (or incompetently) restricted access to needed care to fatten their coffers. With the new MLR law, policymakers hope that insurers:  1) increase medical care costs, or 2) decrease administrative expenses (including marketing costs and profits). 

Instead what will probably happen is that health insurers will increase premiums (while keeping administrative costs the same). This will make the MLR floor easier to obtain. They could also lobby to categorize some administrative costs (call centers, etc.) as medical care. Lobbyists are trying to convince CMS of this now. Worst case scenario, the health insurers will engage in unethical/fraudulent shell games to meet the MLR floor requirement.

Instead of a basic MLR floor, regulators should hold health insurers accountable for measurable quality standards, such as morbidity and mortality. The MLR rule sets a minimum for spending of the premium with medical providers, but, as Robinson (1997) pointed out, “Neither premiums nor expenditures by themselves indicate quality of care.”

Health insurance and healthcare delivery are complicated. In my opinion, the regulating MLR is an over-simplified solution that will not effectively reduce costs or improve quality of our healthcare system. But perhaps that policy horse has already left the barn.

The ACO Paradox

As of today, we are still awaiting the CMS regulations on Health Reform’s Accountable Care Organizations. I’ve made a couple of postings on ACOs in the past. I reiterated the importance of process measures in pay-for-performance schemes, and I scoffed at investor-owned hospitals’ proposal of cherry-picking patients for ACO membership. Here, I’d like to discuss what I’ll call the ACO Paradox.

There are two main components of the Medicare pay-for-performance program (Sec. 3022) for qualified Accountable Care Organizations (ACO), according to the Health Reform Law. First, the Medicare Shared Savings program is designed to reward ACOs for decreasing Medicare fee-for-service costs of their assigned members. If the ACO’s adjusted costs for Medicare parts A & B in the measurement year are less than the benchmark (a 3 year per-beneficiary average), then the ACO and Medicare will share in that cost difference. Second, to qualify for Shared Savings bonus, the ACO must meet quality performance standards (soon to be proposed by CMS).

It makes sense for Medicare to join the costs savings with improved quality. Both are key aims of Health Reform (expanded coverage being another). However, will the ACOs be able to achieve both cost savings and improved quality simultaneously in the measurement year?

The conventional wisdom in healthcare states that increased quality of care will save money in the short-term by reducing high-cost services, such as inpatient care. However, some QOC measures have been shown to be statistically related to increased near-term healthcare costs. For example, a recent study found that improved adherence to antiretroviral regimens used to fight HIV was associated with an increase in total medical costs. Riegal et al. (2000) have posited that the positive relationship of QOC measures to increased near-term future costs is probably due to improved access to needed services.

This isn’t to say that cost savings are not possible with improved quality. It may be that some measures do actually save in the short-term, while others do not. Or maybe it is that some (or all) quality measures save money in the long-term – 5, 10 or 20 years in the future.

So, what if the quality measures that CMS chooses are actually related to increased costs in the measurement period? What if some quality measures are not worth the investment for ACOs? In other words, will there be any Medicare cost savings to be shared in the short-term? Is there an ACO Paradox?

Perhaps Medicare should consider extending the ACO quality measurement and shared savings period to reflect the near-term investment required to produce decreased healthcare cost in the long-term.

Investor-owned hospitals want to cherry pick patients for the Medicare ACO program

Medicare pay-for-performance is a part of the Patient Protection and Affordable Care Act (PPACA) legislation. The health reform law requires the Centers for Medicare and Medicaid (CMS) to create entities that will share in any cost savings for treating Medicare patients. The bonuses would be distributed to these “Accountable Care Organizations” based on healthcare quality performance criteria, such as reduced hospitalization infections.

In a letter to CMS regarding PPACA’s Accountable Care Organizations, the Federation of American Hospitals recommended that ACOs be able to “bring to the agency a list of at least 5,000 Medicare patients that they wish to be assigned to their ACOs.” Among the reasons for this recommendation, FAH writes, is that providers know better than CMS which patients “incongruously bounce from provider to provider, irrespective of the ACO physician’s recommendations (i.e., not a good ACO fit if the patient is unable or unwilling to follow the physician’s advice or remain within the ACO provider network).”

Essentially, FAH, an industry group for a 1,000 investor-owned hospitals, is recommending that they “cherry pick” the best patients. In other words, FAH wants CMS to allow ACOs to behave in ways that lower the likelihood of attracting the unprofitable members.

In my posting last month, I discussed the concept of “cherry picking” in terms of managed care. (This problem is also called risk selection, selection bias or “cream skimming.”) This is a significant problem for the insurance marketplace for which risk-adjusted capitation payment methods are being employed.

With risk selection in mind, CMS must be careful when promulgating the final ACO regulations so as to not inadvertently encourage the discrimination of certain patients (mentally ill, minorities, transient, etc.) that may not score well in performance assessment schemes.

It has already been shown that incentive-based programs exacerbate risk selection problems. Evidence from Mehta et al. (2008) on hospital incentive-based programs studies shows that case mix can exert powerful influence on performance rankings. For example, one study by Karve, et al. (2008) showed that the proportion of African-Americans patients treated by a hospital was inversely associated with performance for certain process-related performance measures. Blatant risk selection policies for the ACO program will make this problem even worse.

If CMS accepts FAHs recommendation for the ACO pay-for-performance scheme, there will be negative unintended consequences.  In my earlier posting about ACOs, I commented that the CMS staff responsible regulations for the ACO program may have an anti-managed care bias. Even if the regulators at CMS dislike managed care, surely they’re aware of the “cherry picking” problem. Hopefully, they will view FAHs recommendation with skepticism.