2. Managed Care / Technology © Allen C. Goodman, 2013

3. Why? We’ve talked about insurance and technology … and costs. Managed care analysis combines some of this. We’ll spend a little bit of time on this here. Dr. Jensen spends considerably more time. Many suppose that insurance necessarily leads to higher costs and perhaps to waste. “Managed care” may address some of these problems. Networks of providers, including Health Maintenance Organizations (HMOs), Preferred Provider Organizations (PPOs), and Individual Practice Organizations (IPOs), are means to restore competition to the health care sector, and to control expanding health care costs.

4. Fee for Service Remuneration Under FFS, provider both provides health care and advises the consumer on how much is needed. It would seem that the consumer’s imperfect information about medicine, when combined with FFS remuneration, would provide the incentives for substantial overconsumption, like SID Organizational form of MCOs would seem to eliminate over- consumption incentives and replace them with cost-control incentives, and possibly incentives toward underconsumption.

5. Managed Care Most generally, analysts speak of an organized delivery system as a –network of organizations (for example, hospitals, physicians, clinics, hospices) that provides or arranges to provide –a coordinated continuum (from well-care to emergency surgery) of services, –to a defined population. This system is is tied together by its clinical (it must TREAT them) and fiscal (it must FINANCE the treatment) accountability for the defined population. Organized delivery system is often defined by its association with an insurance product.

6. Managed Care Shortell and his colleagues view the key feature of managed care as the provision of care to a defined number of enrollees at a capitated, or fixed, rate per member per month. As a result, cost centers such as hospitals, physician groups, clinics, and nursing homes, must be managed under a fixed budget. Under traditional fee-for-service, since cost centers generate revenue, more volume means more profit. Under managed care, more volume means less profit.

7. How good are the information systems? How well do they communicate with each other Some point to the VA as a prime example of one that works!. Shortell et al

8. Some key aspects In Figure 12.1 (from FGS/4), information systems are the hub of the wheel. Large health centers have budgeted hundreds of millions of dollars to integrate systems that were often developed separately, and almost never “talk to each other.” Information is still a SERIOUS impediment – just think about it.

9. Managed care incentives: –keeping people well by emphasizing prevention and health promotion practices, and when people become sick, –by treating them at the most cost-effective (least cost per unit care) location in the continuum of care. There are also incentives to underuse services, and this may be harmful to patients. Through a more centralized management of services, goal is to provide additional quality-enhancing features for a given price, or to provide a given set of quality attributes or outcomes for a lower price. The primary provider has a paramount role as the “gatekeeper” to further, and more expensive, services. Managed Care

10. Point of Service (POS)Fee for Service (FFS) Health Maintenance Organization (HMO) Preferred Provider Organization (PPO) Organizational Structures Gatekeeper Provider Network NoYes No Yes Least Restrictive Most Restrictive

11. Managed Care Plans Who’s buying what? In particular, who’s buying HDHP?

12. Growth of HSA/HDAP Enrollment Source: www.ahip.org/HSA2012/

13. HAS/HDAP Covered Lives Source: www.ahip.org/HSA2012/ Biggest % increases are in large group purchases.

14. Source: Employer Health Benefits – 2012, P. 15 SO = Savings Option

15. Source: Employer Health Benefits – 2012, P. 72

16. Average Deductible by Plan Type – Single Coverage Source: Employer Health Benefits – 2012, P. 116 Deductibles almost doubled Deductibles  only about 20 %

17. Some key aspects De-emphasis of the acute care hospital model. Hospitals provide expensive care, and moving toward cost-effective models necessarily moves away from hospital care. Often, primary care physicians are the gatekeepers of managed care systems, directing patients to appropriate (i.e. cost- effective) treatment settings. Managed care seeks vertical integration of what had been a generally unintegrated system of treatment. Such integration, through coordination of care and improved information, can address health care costs in a manner that would appear to address criteria of economic efficiency. Yet, the integration is costly, and the quality of the resulting care may not match all consumer preferences.

18. Disenrollment Patients and providers face the difficulties of maintaining continuity of care and complete medical records. MCOs face added financial burdens resulting from higher patient recruiting costs, disruption of cash flows, and upward pressure on premiums for continuing members if lower risks are more likely to disenroll. Why emphasize preventive care for a patient who is not likely remain a member, when that care provides the greatest return in the form of averted future treatment costs? Potential loss of patients may influence treatment decisions of FFS providers as well as MCOs, but capitation method of payment to MCOs renders the disenrollment problem particularly important. FFS providers are paid for each unit of care. Aside from uncollectibles, they are not at risk of losing money on services provided currently or in the future.

19. Managed Care, in contrast … In contrast, by integrating insurance with the provision of health care, the MCO receives a fixed payment per enrollee to cover costs in the current period, and over time, for those who remain enrolled. Thus, unlike FFS care, where payment in every period is very likely to cover costs, the MCO must consider the timing of expenditures and the financial losses of overspending on patients who may disenroll. One way for an MCO to “self-insure” against long term losses attributable to disenrollment is to economize on care for those currently enrolled.

20. Some Models - Random Dis-enrollment (1) Begin with an MCO providing coverage (services) x i per enrollee to its insured k i who are representative of the population. MCOs vary by levels k and x, with members changing MCOs due to changing tastes for x, or due to employment-related conditions. Assume that the structure of the firm represents a staff model MCO (in which physicians are firm employees), although the model also applies to other forms of integrated service delivery.

21. Some Models - Random Disenrollment (2) Presume that the MCO recognizes that its pricing decisions may influence its enrollment k. The MCO faces inverse demand curve p = P (k, x), where P k 0. Suppose that there are a large number of consumers with unit demands: - they join this MCO - join another one - purchase from providers in the fee-for-service sector, or - make no purchases at all).

22. Purchasing MCO coverage P (r, x) is the maximum annual premium (including the employer’s contribution) that consumer r is willing to pay for a unit of MCO coverage of quality x. P Number of Consumers P* Consumer Surplus r k

23. Costs C i = the MCO’s total annual cost to provide expected level x i of health services to its k i enrolled members. Total cost C i is related positively to x i, positively to number of person-years of membership k i, and negatively to “health” y of those who happen to be members at the time. With k i as an argument in the cost function we can have either increasing and/or decreasing returns to scale for both k i and x i. x i = services level k i = # of members

24. Costs Assume that health y is positively related to the level of services by all providers. Since at any time in the future, these individuals may be members, average health status y of the population equals  k i x i divided by population K. A constant in the denominator is unimportant in our analysis, so we normalize it as 1; hence y =  k i x i. x i = services level k i = # of members

25. Profit-maximizing MCOs – How much k? Profits for MCO i can be defined as:  i = k i P i (k i, x i ) - C i (k i, x i, y) with C i k > 0, C i x > 0, and C i y < 0 (with C i kk, C i xx, and C i yy all positive). The first decision (optimizing with respect to k i ) leads to the condition:  i /  k i  P (k i, x i ) + k i P i k (k i, x i ) = C i k which states that the premium (i.e. marginal revenue product per enrollee) must equal the marginal cost of the enrollee. Since an individual MCO is small compared to the entire population, the MCO does not recognize its impact through k i on population health status y. x i = services level k i = # of members

26. Profit-maximizing MCOs – How much x? The second decision is how much x i to provide. Again, here through x i, MCO does not recognize impact on population health status y. Thus, for any k i, the maximization is:  i /  x i  k i P x (k i, x i ) = C i x. Marginal revenue of providing one more unit of quality (services) to the entire enrollment k i equals the marginal cost of providing that additional unit. x i = services level k i = # of members

27. The Health Externality Compare the market optimum of marginal revenue equaling marginal cost, to an optimum derived alternatively by a “planner” who is aware of the externality of each MCO’s level of x on health y, or by an entrepreneur seeking to maximize the joint profits of a group of two or more MCOs. Consider a situation with two firms (can generalize to n firms). The entrepreneur seeks to maximize total profits T, or with appropriate substitutions: T =  1 +  2 = k 1 P 1 (k 1, x 1 ) + k 2 P 2 (k 2, x 2 ) - C 1 (k 1, x 1, y) - C 2 (k 2, x 2, y)

28. To maximize profits, the entrepreneur must consider the impacts of health services x 1 and x 2 on the health of others elsewhere. Differentiating with respect to x 1 and x 2 yields: k 1 P 1 x (k 1, x 1 ) = C 1 x + k 1 (C 1 y + C 2 y ), and k 2 P 2 x (k 2, x 2 ) = C 2 x + k 2 (C 1 y + C 2 y ). Compare above to  i /  x i  k i P 1 x (k i, x i ) = C i x. but C i y < 0 implies that the social optimum leads to more treatment x i than the competitive monopolistic market. The Health Externality

29. This is easily shown. Without externality, MCO 1 optimizes at point A, giving level x 1 mkt. P Inputs (services) x 1 k 1 P 1 x C1xC1x x 1 mkt A x 1 opt C 1 x + k 1 (C 1 y + C 2 y ) k 1 (C 1 y + C 2 y ) Optimal level of x 1 at point B is x 1 opt, as noted by downward shift in the right hand side by the factor k 1 (C 1 y + C 2 y ), which is unambiguously negative. This indicates inefficiently small level of MCO care x, and by implication substitution of non- MCO and/or non-health care inputs (such as the patient’s own time) for the MCO care.

30. Disenrollment and Treatment Choice Disenrollment can affect the levels of services x i provided. BUT, by distinguishing only among levels of x i, we haven’t distinguished among alternative treatment methods. Consider longer term consequences that potential disenrollment can have on MCO treatment practices. In the presence of expected disenrollment, MCOs will tend to use “low-tech” treatments with smaller up-front costs, even when the present discounted value (PDV) of the costs equals the PDV for “high-tech” treatments.

31. Consider an infinitely-lived MCO that serves overlapping cohorts of customers for either one or two periods. The MCO receives constant revenue per patient each period, so its goal is to minimize costs. Costs are modeled for Cohort 1 entering in Period 1: For “high-tech,” possibly capital-intensive procedures, the MCO incurs costs M 1 in Period 1, and 0 in Period 2. For low-tech, less capital-intensive procedures, the MCO incurs costs m 11 and m 12 in Periods 1 and 2, where m ij refers to cohort i at time j. Disenrollment and Treatment Choice Goal! Minimize Costs

32. For Cohort 1, the MCO would plan “high-tech” procedures if: M 1 < m 11 + m 12 / (1 + r). where r is the appropriate interest rate for discounting. However, suppose that potential MCO members use expenditures M t and m tt as indicators of MCO quality. So the MCO uses high- tech procedures for Cohort 1 if: M 1 < m 11 +(1 - g) m 12 / (1 + r). To maintain revenue, Cohort 2, of size g, must be enrolled in Period 2. The Cohort 2 MCO decision is similar to the Cohort 1 decision if: M 2 < m 22 + (1 - g) m 23 / (1 + r). So, for any two period sequence, the firm chooses the high-tech treatment if: M 1 + M 2 /(1 + r) < m 11 +(1 - g) m 12 / (1+ r) + m 22 /(1 + r) + (1 - g) m 23 /(1+ r) 2. Disenrollment and Treatment Choice

33. Over time, the MCO will be indifferent between high-tech and low-tech options if:  M t (1+r) 1-t =  [m tt (1+r) 1-t + (1-g) m tt+1 (1+r) -t ]. Assuming that all M t = M, and that all m ij = m, expanding both sides, and taking limits as t  , the MCO chooses high- tech if: M < m[1 + (1-g)/(1+r)], and low-tech if: M > m[1 + (1-g)/(1+r)]. Thus the higher the disenrollment rate g, the more important is the disenrollment effect. If g = 0, the MCO faces the standard investment criterion, comparing first period costs with discounted future costs. Its decision here will be economically efficient. Disenrollment and Treatment Choice

34. With the likelihood that increased competition through increased choice will raise g, managed competition and other competitive strategies must be more carefully examined. If g , m [1 + (1-g)/(1+r)] , and continuing care m becomes the more financially viable option even if the PDVs are equal, and even if treatment M is more economically efficient in producing health. In effect, MCOs self-insure against future disenrollment by reducing current costs through (low cost) continuing care rather than high-tech treatment.

35. Potential Issues: Dumping, Creaming, Skimming Dumping – Refusing to treat less healthy patients who might use services in excess of their premiums. Creaming – Seeking to attract more healthy patients who will use services costing less than their premiums. Skimping – Providing less than the optimal quantity of services for any given condition in a given time period. MCO: Probably. FFS: Probably not – Reimbursement will cover. MCO: Probably. FFS: Probably. MCO: Probably. FFS: Probably not – Reimbursement will cover.