Administrators of community mental health centers, at-risk behavioral health plans, and other large public-sector behavioral health providers find themselves struggling to manage the expense of contemporary psychopharmacological agents. Although there is little doubt that these medications are cost-effective (1,2,3), the recent escalation in medication expenses (4,5) puts a significant pinch on mental health care budgets. In Texas, for example, community mental health centers are responsible for paying for medications but do not reap the financial benefits resulting from the ability of contemporary medications to reduce other health care expenses. Therefore other services, particularly rehabilitation services, invariably are reduced to allow these centers to pay for medications. Clearly this reduction in services is problematic for patients suffering from the psychosocial sequelae of chronic mental illness. The reduction may also serve to further escalate medication expenses, because physicians may attempt to treat psychosocial problems with heftier doses of medications.
In this context, many health care plans and large providers have instituted formulary restrictions (6,7), often unilaterally. Because of a significant escalation in the expense of psychoactive medications, our health plan faced similar restrictions either on medications or on the range of services available to our chronically mentally ill patients. We elected to take the issue of pharmacy expenses to our plan stakeholders instead of instituting changes in isolation. Our goal was to develop a strategy that could be implemented by consensus, thus improving its chances for success.
The formulary guideline project was undertaken as part of the NorthSTAR program, a mental health and substance abuse carve-out intended to integrate Medicaid, state general revenue, and federal block grant funds into a single system of public behavioral health care. The program provided services to a seven-county service area including Dallas, Texas. The state contracted with two behavioral health managed care organizations to provide services under a full-risk capitated arrangement.
The primary goal of NorthSTAR was to increase access to behavioral health services. From its inception, the program was successful in attaining that goal. Increased access could have had a dramatic effect on the cost of services, but the goal was to improve access by improving the efficiency of service provision.
In general, the plan met its access goal through the first few months of its implementation. However, several factors led to a dramatic increase in pharmacy expenses, an increase that did not seem amenable to the forces of fiscal efficiency that are typical of managed care.
As barriers to care dissolved with the rollout of NorthSTAR, the number of enrolled members who received pharmaceutical agents grew. In the first full month of the program, one of the behavioral health managed care organizations provided medications to 3,700 patients. Within six months, 4,300 patients were receiving medications each month, a 30 percent annualized rate of increase. Medication costs increased at a 36 percent annualized rate over the same period. The problem was exacerbated by the fixed cost to the health plans of state hospital beds regardless of use, which negated the pharmacoeconomic advantages of some contemporary medications, most notably the new-generation antipsychotics. Given that the potential growth in program membership was open-ended, there was significant concern that pharmacy expenses would continue to increase.
The following four primary factors seemed to drive the increase in pharmacy expenses while other program expenses remained steady or declined.
Inflation in the cost of drugs
Annual prescription drug expenditures in the United States increased from $70 million in 1995 to $90 million in 1998 (1). Although much attention has been given to the impact of new-generation drugs on the increase in the cost of psychotropic medications, almost half of the increase resulted from higher costs of older medications. One-third of the increase was due to inflation in the cost of established agents (1). This upward pressure on overall medication costs was evident in NorthSTAR.
Pharmaceutical manufacturers have turned to advertising directly to patients. With $1 billion spent on this approach in 1997, the industry is now spending more money on direct advertising than it does on traditional medical journal advertising (8,9). The increase in direct advertising may stem in part from the growth of managed care formulary restrictions, which has made it difficult for pharmaceutical manufacturers to influence physicians directly (10). Although it can be argued that advertising improves patient education and leads to consumer empowerment, the ultimate goal of marketing is to increase sales. The continued growth in direct marketing suggests that manufacturers believe it accounts for an increase in prescription drug expenditures.
Providers' poor knowledge about price differences
Similar to the experience of others (10,11,12), our network physicians generally had little knowledge about price differences among medications. In many settings, prescribers learn about relative medication costs only after managed care shifts the financial risk to contracted physician providers (13,14). However, no such incentives were in place in the NorthSTAR managed care pilot program.
Costs isolated from market forces
The final factor in the escalation of medication costs appeared to be their isolation from typical managed care cost-management strategies. When an at-risk managed care organization contracts with acute care facilities, it generally pays about the same price per bed-day at one facility as it does at another. Market factors tend to minimize cost differentials. By comparison, the pharmaceutical industry generally has not been affected by these market factors, because manufacturers hold that they have little leeway to compete on price because of the high cost of bringing a new product to market. In general, payers have been willing to pay a wide range of prices for medications with the same clinical indications but different adverse-effect profiles.
Strategies were developed to address each of the four issues identified above. With regard to price inflation, we decided that the most effective approach would be to share risk with member-patients through copayments. Copayments also can serve to educate members about the cost differentials among similar products. Therefore our initial strategy was to propose a maximum benefit for two classes of medications—antidepressants and new-generation antipsychotics. The benefit would be defined as the cost of an average one-month supply of the least expensive agent in each class. Members would be responsible for the balance of the cost of their chosen medication.
For example, we found that the average cost of a one-month supply of citalopram was the lowest of any nongeneric antidepressant. Our proposal defined this cost as the benefit, with members paying the difference if they chose to use another nongeneric antidepressant. The new-generation antipsychotics had a similar cost differential; the mean daily cost for members who were taking risperidone was significantly less than the cost of the other three agents.
However, administration of a varying copayment would have been difficult. Therefore we revised the proposal to a fixed copayment equal to the difference in the mean cost between a preferred agent and all other agents in the class. This copayment turned out to be $30 a month for antidepressants and $60 a month for new-generation antipsychotics.
With regard to direct-to-consumer marketing, we decided that the best approach would be to market just as aggressively as the pharmaceutical manufacturers do. Our marketing strategy would center on educating our various stakeholders, particularly advocates and consumers. The education would focus on expert consensus on the relative efficacy of medications within a class. Specifically, we proposed to counter the product-differentiation strategy used by the manufacturers of the highest-priced antidepressants and the new-generation antipsychotics by educating about the agents' relative efficacy.
We also planned to focus on the potential impact of medication costs on the range of program services offered. To address the issue of providers' knowledge about medication costs, we initiated an educational process with major physician providers in our network. Because our highest-volume providers were community mental health centers, we believed that our network physicians would be amenable to arguments that addressed the issue from a public-health perspective. In particular, we wanted to focus on our ability to treat more patients with the same number of dollars if certain medications were used in first trials.
The isolation of medication costs from managed care could be addressed by fostering competition among pharmaceutical manufacturers. We proposed eliciting competition by naming preferred brands of new-generation antipsychotics and antidepressants. We discussed this idea with psychopharmacology experts to ensure that it would not adversely affect quality of care. We then told manufacturers' representatives that we would name a preferred brand in these two categories on the basis of the mean monthly expense of treatment.
The use of preferred brands required a mechanism to allow physicians to deviate from the guidelines when doing so was clinically indicated. To enhance the integrity of this approach and also to improve quality of care, we proposed to tie authorization of preferred-brand "waivers" to clinical practice guidelines. Thus, as has been done in other disciplines (14,15), we proposed using the Texas Medication Algorithms (16,17) to guide approval of medication changes.
In keeping with our overall strategy, we developed proposed formulary guidelines on the basis of the strategies described above and took them to our stakeholders for their feedback. We held a series of meetings with our state contractor, our major physician providers, our consumer advisory group, local advocates, and representatives of major pharmaceutical manufacturers. In addition, we distributed our proposal through the local behavioral health community by using handouts and electronic mail lists.
In each of these meetings we stressed our desire to develop a consensus approach to managing pharmaceutical costs that involved all program stakeholders. We then explained our initial proposal and the rationale behind it. Finally, we elicited feedback on the proposed changes to ensure that the final formulary guidelines would be acceptable to all.
In meetings with our state contract managers, we were asked to lower the proposed copayment to $20 per prescription, which would be consistent with the copayment that state employees were currently paying for their own medications. The impact of this requirement was to shift a significant proportion of the cost differential between preferred and other brands back to the managed care organization.
In meetings with physician providers, we were surprised at the degree of consensus on the concept of preferred brands. For the new-generation antipsychotics, the choice of risperidone as the preferred agent drew virtually no criticism. The choice of citalopram as the preferred antidepressant drew some resistance, and the recommendation was made that a second antidepressant be added. On the basis of relative cost, we suggested that bupropion be the second preferred agent, and this choice was generally accepted.
The initial response from consumers and community advocates was quite muted. One local advocate presumed that this was due to "shock that our opinion was even requested." Their primary concern was the fate of patients who were already receiving medications through the program and the status of their copayment requirements. As initially proposed, the plan could have required patients who were currently receiving higher-cost medications to either begin a copayment or switch to the preferred brands. There was understandable resistance to this notion, so the plan was modified to "grandfather in" patients who were stable with medications, thus greatly reducing the proposal's ability to reduce medication expenses. We also developed a process for waiving copayments after the failure of preferred-brand medications in trials using Texas Medication Algorithm Project guidelines.
The pharmaceutical industry representatives were surprisingly cooperative. Most representatives understood the rationale behind our proposal. In a spirit of business competition, most set about developing marketing strategies to counter our decision.
All of the feedback was incorporated into a final draft document. In summary, the consensus guidelines had these primary features: a preferred antipsychotic (risperidone), two preferred antidepressants (citalopram and bupropion), a copayment of $20 for new starts on other nongeneric antidepressants and new-generation antipsychotics, and a process for appealing copayments when preferred-brand medications failed in trials using Texas Medication Algorithm Project guidelines.
The final draft was presented at a stakeholder meeting arranged by the local Mental Health Association, which included advocacy group representatives, local mental health agency leaders, private-sector providers, community mental health center leaders, and consumers. Given the level of trust that the guideline development process had fostered, the meeting participants approved the final draft quickly. One community advocacy leader then steered the topic away from the formulary guidelines and toward the core problem—the lack of adequate funding for an unrestricted formulary. The group began to make contingency plans to lobby at the state level for more medication funding.
The new consensus formulary guidelines were implemented about six weeks after the final meeting. However, the impact of our process in terms of engaging stakeholders was apparent long before the actual implementation date. Education of physicians led to an immediate increase in the use of less expensive antidepressants: citalopram prescriptions increased by 40 percent in the month before implementation of the guidelines.
A comparison of medication users and expenses for December 1999 (before the stakeholder discussions began) and June 2000 (two and one-half months into implementation of the guidelines) also showed the impact of the plan. The total number of medication users increased by 3 percent over this period, but overall medication expenses decreased by 3 percent. The number of users of selective serotonin reuptake inhibitors increased by 11 percent, yet the costs of these agents were unchanged. The number of users of new-generation antipsychotics increased by 3 percent, and total expenses for new-generation antipsychotics decreased by 6 percent.
In the 12 weeks after implementation of the new guidelines, very few complaints were received from consumers or families. The most frustrating process for physician providers was the lag time for approval of new medication starts. Most facilities offered medication samples during the lag time.
Providers' resistance to the new formulary guidelines came primarily from nonnetwork, private-sector physicians, who had not been involved in the stakeholder meetings. These complaints seemed to stem from a concern that the lower level of physician autonomy under the guidelines would be "infectious" to other community managed care plans. Complaints were directed to the local medical society and to our state contractor.
Noting several studies that showed that the initial medication expense does not represent the overall cost of treatment (18,19,20,21,22), several pharmaceutical manufacturers initiated projects to demonstrate the relative cost-efficiency of their products by using complex pharmacoeconomic models. Some manufacturers were interested in rebidding for designation of preferred-brand status. Most manufacturers enhanced their marketing to physician providers as a counter to our provider education process. One of the major manufacturers also significantly increased its marketing to advocacy groups and its lobbying to state officials.
The public-sector formulary guideline development and implementation process presented in this article has two primary advantages. First, education and marketing can be used to engage stakeholders in the management of pharmacy expenses. Second, such engagement can be achieved without the traditional adversarial relationships of managed care systems. The impact of provider education was demonstrated by the change in antidepressant prescribing patterns that occurred before the consensus formulary was implemented. Finally, the paucity of complaints from stakeholders after implementation of the guidelines points to the value of the engagement process in positioning managed care.
The implementation process described here does have some drawbacks. First, it consumes a considerable amount of stakeholder time and therefore is fairly expensive. However, in the context of diminishing resources for mental health care, the investment of time and energy into collective ownership of difficult issues such as the balance of access and quality may actually enhance the overall efficiency of the system and also solidify advocacy efforts and their effectiveness. Second, we lack data on the impact of the guidelines on quality of care. The pharmaceutical representatives may be able to assist in providing some of these data in the future.
Clearly this process would not have been necessary if not for the primary frustration for all of the stakeholders—the lack of adequate funding for public mental health care. For that reason the final meeting with the stakeholders, in which energy was refocused onto enhanced advocacy efforts, was especially gratifying.
When this work was done, Dr. Baker was medical director at Magellan Behavioral Health of Texas in Dallas. He is now medical director at Dallas MetroCare Services, 1380 River Bend Drive, Dallas, Texas 75247 (e-mail, firstname.lastname@example.org).