Depression is one of the most common and disabling conditions on the planet and despite the availability of efficacious drugs and psychotherapy there is little evidence that the burden of depression is decreasing.1,2 Because nearly all patients with depression who seek medical care for depression are seen in primary care, this clinical setting has been the target of intense clinical research to improve recognition and treatment.3 Unfortunately, treatment for depression in primary care is, and always has been, characterized by the provision of psychoactive drugs and unstructured counseling for nonspecific emotional symptoms.4 Under use of prescription medications by primary care physicians appears to be an unlikely explanation for the continued burden of depression. The United States now spends more than $12 billion annually on approximately 190 million prescriptions for antidepressant medications.5 Most of these drugs are prescribed by primary care physicians and patterns of prescriptions do not comply with a model that suggests a specific drug for a specific psychiatric illness.6,7 How do we resolve the paradox that antidepressant drugs work and we frequently prescribed them in primary care but patients do not seem to get better?
Looking through the PRISM of a quarter century of Depression in Primary Care research, we recall our DART out of PORT with our Partners in Care on our QuEST to identify a PATHWAY to improve primary care physicians' RESPECT for the complex care needs of patients with depression. The TIDES now seem to suggest the PROSPECT that we might find PEARLS beneath the WAVES and therefore make a real IMPACT on the SADHARTS of primary care patients.8–23 On this voyage, we have encountered patient educational campaigns, patient kiosks, and direct-to-consumer advertising, physician sticky note reminders, computer reminders, and computer-based decision support, population-based screeners and primers, psychiatrists and academic detailing in primary care, nurses in the clinic and in the home, nurse on the telephone and on the web, and primary care doctors as villains, fools, gatekeepers, opinion-leaders, and team members. If you are a primary care physician and you do not know the story behind these acronyms and interventions, you should know that a wealth of clinical research certifies that major depression is treatable in the broad range of primary care patients.
If you are a primary care physician, you should also know that this rich literature shows that writing a prescription for an antidepressant drug and handing it to a patient who seems to be sad or needy is not treatment for major depression. Patients won't take the drug long enough for it to be effective,24 many won't have an illness that responds to the drug anyway, and the patient won't have much confidence in you, especially when he or she discovers the side effects and price of the drug. A simple prescription might help you get to the next patient sooner, but it won't help the one you are with. If you are a mental health professional waiting in your office for referrals, you are about as much in the game as a spectator in the upper deck of your home stadium (during an away game). If you are a health system leader you should know that depression awareness flyers posted in the hospital cafeteria, continuing medical educational seminars, leather-bound practice guideline manuals, and feedback of screening scores are about as effective as handing out lollipops. There is no quick inexpensive fix to the problem of depression in primary care, and there are many clinical trials to prove it.
Both antidepressants and psychotherapy are effective treatment for the specific illness of major depressive disorder when they are used in conjunction with a system of care that includes patient activation, close monitoring of patient response, appropriate adjustment in treatment based on the patient's response, patient's access to support beyond the office visit, appropriate reimbursement, access to specialty care, and an empathetic care environment both inside and outside the health care system. When such a system is available, 50% to 70% of patients will experience improvement. If this is the story told by a quarter century of clinical research, how do we implement these lessons across an entire nation? As researchers, clinicians, and educators, we are accustomed to making the scientific case and the patient case. We do not make the business case. Or do we?
Three papers in this month's Journal of General Internal Medicine extend our knowledge of treatment for depression in primary care by moving progressively into the realm of effectiveness, dissemination, and implementation science.25–27 It is wonderful to note that each of these papers begins with the assumption that research already demonstrates the effectiveness of treatment for depression in primary care. There is a wealth of methodologic innovation in these studies, and each pushes the envelope in areas such as community-based laboratories, study design, outcomes measures, or analytic approach. However, each study also seems to be pushing the envelope in terms of building a business case for depression treatment. This is very thin ice, and it is ice that mental health policy advocates have already fallen through more than once (e.g., the Holy Grail search for a cost-offset effect).28 There is danger in building a business case for the treatment of depression around the possibility that it improves outcomes for another disease that we really care about—for example, heart disease, diabetes, or cancer. There is danger in building a case for treatment of depression because it does not cost too much. How much is too much, and to whom and compared to what? There is even danger in suggesting to health care systems that it would not take much to implement quality improvement for depression. Research demonstrates that treating depression is complicated and requires fundamental system redesign. None of the authors of these 3 papers directly suggest any of these business case strategies, but the language of the business case is growing in the mental health services literature.
Mental health services researchers at the University of Washington demonstrated that depression is treatable even among patients with comorbid and symptomatic chronic medical conditions such as diabetes.25 This is the latest in a series of sustained work by this team that applies the principles of the chronic care model, including stepped care and collaborative care, in the real world of primary care. This work highlights the dividends of a large health care system investing in a primary care laboratory and the fruits of bringing together psychiatric and primary care expertise, among others. This work also demonstrates that depression is treatable even among patients with comorbid conditions. However, there is currently little evidence that successful treatment of depression provides a bonus in improved self-management or outcomes for comorbid conditions. We do not have to rationalize expending resources on depression because it will improve outcomes for some other chronic condition. Or do we?
Bosmans et al.26 randomized 34 general practices in the Netherlands to evaluate the cost-effectiveness of a disease management program among elderly primary care patients. This study tackles the difficult issue of how a large health system implements evidence-based quality improvement within an environment of competing demands. The study also moves further along the path of randomizing primary care practices using a cluster-randomized trial and including both cost-effectiveness and cost-utility as outcomes. Such studies are important because national health systems are increasingly making resource allocation decisions on a currency of quality-adjusted life years or QALYs. Notably, the intervention tested by the Dutch team combined screening, diagnosis, and decision-support but did not include additional team members or outcome tracking support, and there was minimal structured involvement of mental health specialists. Despite the substantial system redesign needed for this scaled down model, there is no evidence from this study or prior work that this intensity of practice redesign is sufficient to change the process or outcomes of care. As stated by the authors: “we chose a pragmatic design, meaning that we tried to resemble everyday clinical practice.” Such decisions have typically resulted in a dilution of the multicomponent care models to a level of ineffectiveness. How much of the complex model does a system really have to implement in order to get results? So far the answer is: you need to implement the whole model. It is tempting to try to prove cost-effectiveness by reducing the cost of the intervention. However, we do not have to prove that depression delivers substantially more QALYs per dollar than treatment for other common medical problems. Or do we?
The study by Rubenstein et al.27 represents the state-of-the-art in implementation science. Like a wonderful piece of art, this paper is to be studied and enjoyed from multiple angles ranging from the intricacies of the methods to the insights of the discussion to the call for further debate. The 3-dimensional Figure 1 that depicts the study design beautifully captures the complexity and creativity of this work. If you are a student of health services research, this study will fill you with “both encouragement and caution.” The study is encouraging because we have a picture of the future of implementation science. The study signals caution because there is an enormous amount of research expertise, experience, and infrastructure behind this study. It also emits caution because the implementation of such a study requires a complex interplay of science with political art at the level of individuals and organizations. Essentially, every sentence of the methods is a testimony to complexity and hard work and human behavior. This type of work moves the researcher out of the laboratory and into the boardroom. Capturing the essence of the trenches, the authors report that: “the mental health leadership for the 2 VA practices openly opposed primary care treatment of depression.” Almost all depression treatment takes place in primary care—we do not have to make that argument yet again. Or do we?
In the intervention group, the UCLA research team helped the health system leadership identify priority quality improvement practices and provided the tools and strategies for implementing the change. Many of the targeted practices decided to implement low cost, low-hassle quality improvement strategies even when presented with evidence that such strategies are ineffective when not part of a larger package of interventions. Moreover, several of the practices could not even implement the new practices their own leadership identified as priorities. Consistent with the findings of Bosmans et al., the low-cost, low-hassle approach does not seem to change process or outcomes of care for depression. Among the practices that did not take the easy route, Rubenstein et al. found important and significant improvements in patient outcomes even at the practice level. It is important to highlight that this is not a study of whether depression is treatable. This is a study that recognizes that changing practice for the care of depression in primary care is difficult, and we need research to help us understand how we can facilitate practice change. Rubenstein et al. show that you can lead a horse to water and make them drink, but it is not easy.
Research on depression in primary care has come a very long way in the past quarter century. Mental health advocates have been pounding out the message that depression is treatable, and there is excellent evidence to support this message. However, there is a new message that depression treatment is not easy. Depression treatment models are complex, they cost money, and they cannot be provided without substantial practice redesign in primary care. An important minority of patients with depression will need specialist care and we need to understand how to get these specialists in the game. It seems clear that science is not enough. There is clearly a cultural and political barrier to effective care for depression. Rubenstein et al. appropriately suggest that researchers “consider which evaluation designs and reporting standards should be used in quality improvement interventions.” Measures of statistical significance rarely move the boardroom discussions forward. The language of science and the language of business are not interchangeable. The papers in this month's JGIM show a progressive movement of our research infrastructure out of academic silos and into the community. In addressing all the “Or do we?” questions, the answer is yes. We have already rededicated some of our finest scientists, best laboratories, and limited research funding to the business case. We should do so with enthusiasm and caution and with both eyes open.