Antidepressant medications were the third most commonly prescribed class of drugs in the United States in 2008 (1). Their use has grown steadily over recent decades, and continued growth is likely. The reasons are that the lifetime prevalence of major depressive disorder is high, ranging between 15% and 17%; its usual symptom onset occurs early in life, during teen or young adult years; it features an episodic, recurrent lifetime course that worsens over time and becomes chronic when untreated; and it generates a staggering 10%—11% of all medical burdens and disabilities (2, 3). Yet, at a maximum, only about 30%—40% of the millions with major depressive disorder currently receive adequate treatment. The unmet demand is huge.

Clinical depression also is costly. Total annual expenses were estimated a decade ago to exceed $81 billion (4) and almost certainly are higher now. In the United States, only cardiovascular disorders appear to generate greater costs.

Antidepressant medication costs represent only a relatively small fraction of the disease's total expenses, but steady increases in medication costs have led to cost-control efforts targeting drugs for the past several decades. Despite years of searching for good strategies, effective pharmacoeconomic models remain elusive. Furthermore, large studies linking clinical outcomes to cost data remain almost nonexistent. This means that there is no prevailing agreement about how to best control or reduce antidepressant medication expenditures while successfully treating and maintaining wellness for the millions with major depressive disorder.

The study by Mark et al. (5) in this issue of the Journal evaluates a current, prevalent strategy known as step therapy, defined as "a type of pharmaceutical benefit design that requires that patients try certain specified medications (typically generic medications), prior to using alternative, more expensive medications within the same medication class." Step therapy permits different types of medications to be tried in subsequent stages rather than rigidly requiring generic substitutes for the same medication in branded form. Previous studies have reported that antidepressant step therapy increases use of generic medications while reducing pharmaceutical expenditures (6). This study's foci are on pharmacy, medical utilization, and spending, not on clinical outcomes or long-term wellness; this is an important point.

Articles addressing healthcare cost-reduction strategies are unlikely to be favorite reading for most clinicians, but this topic is relevant. Why? First, the authors' findings and discussion require us to consider whether step therapy may inadvertently be activating the "law of unintended consequences" (what happens when a simple system tries to regulate a complex system; changing one variable may create a cascade that produces unwanted results) and launching a cascade that actually worsens both overall well-being and cost consequences. Second, the questions generated in the article of how to control costs while producing good outcomes serve as a virtual mandate to us to develop alternative, integrated models that incorporate rather than ignore clinical measures when seeking to achieve cost control.

Specifically, to address several gaps in available step therapy data, the Mark et al. study evaluates associations between implementation of step therapy on medication and nonmedication medical care utilization (changes in prescription drugs and out-patient, inpatient, and emergency room visits) and spending. Using data from over 60 employers, more than 15,000 antidepressant users enrolled in step therapy were compared with more than 45,000 antidepressant users in comparison plans. Expenditures included the total amount reimbursed to providers from all sources of payment.

As observed in other studies, the numbers of days for which antidepressants were supplied and medication costs decreased after step therapy was implemented. However, step therapy appeared to reduce pharmaceutical costs at the unintended "cost" of driving up emergency and inpatient costs. These unintended "costs" have the potential over time to become far greater contributors to total costs of treatment. For example, one tragic near-death suicide attempt in an untreated patient that results in an emergency room visit, prolonged intensive care, and lengthy rehabilitation may overwhelm projected savings from step substitutions for many patients.

Mark et al. review possible reasons for their finding that alternative costs seemingly developed. These include the step process having deterred some patients from filling medication prescriptions at all. Motheral et al. (7) reported that 17% of step therapy members who had experienced denials for selective serotonin reuptake inhibitors, proton pump inhibitors, and nonsteroidal anti-inflammatory medications received no medications at all; another 10% only received a sample or an over-the-counter alternative. Medicaid patients receiving antipsychotics in a step therapy program were also noted to have greater medication discontinuation (8). As an action, they suggested that payers and patients improve strategies to address denials and other barriers to step 2 medications.

Arguably, a more important question is raised by the findings of Mark et al.—a question that suggests we need a new model of cost control. Could the law of unintended consequences ominously be interfering with antidepressant treatment effectiveness, for what is clearly a chronic illness, and, in the process, sustaining or further accelerating prevailing high total expenses?

Despite the importance of the question, we do not know the answer. We do know from an array of studies, however, that major depressive disorder tends to be lifetime, episodic, and recurrent (9, 10) and that three integrated clinical concepts are vital if we are to obtain both good outcomes and cost containment for those with the disorder. These are attainment of wellness, maintenance of wellness, and high adherence. Over one's lifetime, these may be the only goals likely to meaningfully reduce both clinical morbidity and costs while improving productivity. Unfortunately, none of the three are easily achieved, perhaps inviting the search for "simple" solutions that activate the dreaded law of unintended consequences.

Sequenced Treatment Alternatives to Relieve Depression (STAR*D) (11) data illustrate the complex treatment-resistant and chronic nature of clinical depression. Attainment of wellness (remission) could not be achieved in approximately 30%, even after four levels of treatment with different drug classes, such as those used in step therapy. With each unsuccessful effort, treatment resistance, recurrences, and chronicity became more likely and costs rose. Second, antidepressant treatment and other treatment approaches, such as cognitive-behavioral therapy, to maintain wellness for those with prior repeated episodes of major depressive disorder greatly improve longitudinal functioning (9, 10). For those with prior episodes, recurrences of major depressive disorder are virtually predictable for most individuals if antidepressant treatments are discontinued. Finally, high adherence (compliance) is crucial for both attainment and maintenance of wellness (12). Stated simply, for those that are fortunate enough to be receiving an antidepressant that is effective, staying with the treatment is required for the benefits to persist. Similar lessons were learned for diabetes medications decades ago.

It has been known for some time that overall healthcare costs receive less priority from any given segment of healthcare payers and providers than concern about their particular part of overall healthcare costs. When actions are taken, however, without considering the larger, complex picture, unintended consequences are likely.

The sophisticated study by Mark et al. calls for future action items, but a suggested list will sound different. A paradigm shift in our models is needed: