Factors Associated With Primary Care Clinicians’ Choice of a Watchful Waiting Approach to Managing Depression Lisa S. Meredith, Ph.D. Wendy J. Y. Cheng, Ph.D. Scot C. Hickey, M.A. Megan Dwight-Johnson, M.D., M.P.H.
Objectives: Watchful waiting to manage depression in primary care may be an appropriate management approach for some patients who present with less severe depression. This study examined factors associated with primary care clinicians’ choice of a watchful waiting approach to care management for depression. Methods: Secondary data were analyzed from Partners in Care, which examined dissemination of best practices for depression in primary care. Primary care clinicians’ decisions regarding watchful waiting were examined by using the baseline survey data from Partners in Care completed by clinicians and patients from February 1996 to March 1997. Participants were 167 primary care clinicians from 46 practices of seven managed care organizations across the United States and their 1,187 patients with depression. Primary care clinicians’ proclivity for watchful waiting was examined by using a brief scenario describing a patient with major depressive disorder. Results: Thirty-four clinicians (20 percent) reported a strong proclivity to use watchful waiting for the patient in the scenario. The proclivity was significantly associated with clinicians’ reports of the proportion of their actual patients with whom they used this approach. Clinicians were significantly more likely to choose watchful waiting for their actual patients if they had more psychotherapy knowledge (p=.035) or perceived that the need to treat the patient’s medical illness was more important than the need to treat his or her mental illness (p=.046) and were less likely to choose a watchful waiting approach if they perceived the lack of availability of mental health professionals as a barrier (p=.050). Conclusions: Primary care clinicians’ knowledge of treatment and perception of barriers influence their proclivity for watchful waiting. Clinician education to promote appropriate use of watchful waiting on the basis of clinical need is recommended. (Psychiatric Services 58:72–78, 2007)
Dr. Meredith and Mr. Hickey are affiliated with the RAND Corporation, 1776 Main St., Santa Monica, CA 90407-2138 (e-mail:
[email protected]). Dr. Cheng is with the Department of Educational Psychology, California State University, East Bay. Dr. Dwight-Johnson is with the Department of Psychiatry and Behavioral Sciences, University of Washington, Seattle. 72
PSYCHIATRIC SERVICES
U
se of watchful waiting, defined as assessment with scheduled follow-up in primary care but no active medication or psychotherapy treatment to manage depression, remains debatable because evidence has provided mixed results. Some research suggests that watchful waiting may be an appropriate management approach for patients who present with less severe depressive disorders for several reasons (1–3). Many patients with minor depression recover spontaneously within two to four weeks and thus may not need treatment if they do not present with severe symptoms, substantial functional impairment, or long illness duration. Even with specific treatment, these patients may fare no better than they would if they received supportive care or watchful waiting (4). Additionally, many patients with depressive symptoms related to normal bereavement are good candidates for watchful waiting because symptoms often subside within two months. A watchful waiting approach may also promote patient-centered care by allowing for patient preference when there is no definitive treatment choice. Other studies and evidence-based guidelines, however, suggest that pharmacotherapy or structured forms of psychotherapy should be used as a first-line treatment for major depressive disorder (1,5). Illuminating the factors associated with primary care clinicians’ use of watchful waiting
♦ ps.psychiatryonline.org ♦ January 2007 Vol. 58 No. 1
may address gaps in current depression management guidelines and clinician education. For many primary care patients with depression, their depression goes unrecognized, and even when their depression is recognized, they do not receive evidence-based care (6,7). Among those offered treatment, nonadherence rates are still high: as many as one-third of patients discontinue medications after one month, and nearly half discontinue medications after three months (8). Thus it is difficult to know whether clinicians choose watchful waiting to benefit the patient or to cope with numerous barriers, including patient reluctance to acknowledge symptoms or initiate and maintain treatment (9), and practical barriers, such as insufficient insurance coverage and lack of transportation, child care, or flexible work schedules (10). Barriers may also stem from the primary care clinician’s competing demands—that is, addressing patients with multiple medical problems, time constraints, and difficulty coordinating care with mental health specialists (11). Clinicians may also have limited training, knowledge, and skills for the detection and management of depression (12). Clinicians may be uncomfortable treating depression or believe that it poses an unmanageable burden to their practice (13). A consistent deficiency in primary care management of depression is the lack of resources to provide adequate followup, monitor symptom progression, and adjust the treatment regimen. These barriers are interlinked with structure and organization of primary care practice (14), including clinician financial incentives that may drive treatment choice (15). On the basis of social psychological theories about decision making—for example, social-cognitive theory (16), the theory of planned behavior (17,18), and the health belief model (19,20)—we hypothesized that clinicians who perceive no barriers to treatment and who have more knowledge and confidence will be more likely to use a watchful waiting approach for treating depression. Our framework incorporates individual clinician and clinician-level patient PSYCHIATRIC SERVICES
factors—that is, demographic characteristics and depression severity for each clinician’s patients in the study—and system factors—such as organizational integration between primary care and behavioral health services (14) in influencing clinicians’ tendency to use a watchful waiting approach. In this article, we identify patient, clinician, and system factors that are associated with primary care clinicians’ tendency to use a watchful waiting approach; data were from the
Many patients with minor depression recover spontaneously within two to four weeks and thus may not need treatment if they do not present with severe symptoms, substantial functional impairment, or long duration of illness.
Partners in Care (PIC) study (21,22). A better understanding of factors associated with a higher tendency for clinicians to use watchful waiting to manage their patients with depression can provide information about clinician values regarding depression care management and can identify and address barriers to optimal depression care in primary care settings.
Methods Participants We analyzed data from 167 primary care clinicians (92 percent of 181
♦ ps.psychiatryonline.org ♦ January 2007 Vol. 58 No. 1
clinicians eligible for this study) from 46 practices of seven managed care organizations (MCOs) across the United States who participated in the PIC study and from their 1,187 patients with depression. Within each MCO, we categorized clinical units (or clinics), which could be a single clinic, a cluster of small clinics, or a clinical care team within a large clinic. We then grouped these clinics into blocks of three within each MCO, matching on the primary care clinician’s specialty mix; the patient’s demographic characteristics, including racial and ethnic characteristics; and level of behavioral health integration. Within each block of matched clinics, we randomly assigned one clinic each to two quality improvement programs (enhanced medication and enhanced psychotherapy) and usual care. Additional information about the study design is available on the PIC Web site (www.rand.org/ health/projects/pic). Measures Data were collected by sending selfadministered mail-back surveys to clinicians (February 1996 through March 1996) and patients (May 1996 through September 1996) before the study interventions were fully under way (June 1996). The surveys consisted of batteries that have been evaluated previously for reliability and validity and batteries developed specifically for this study. The following scenario was used to depict a patient with symptoms of major depressive disorder presenting for treatment: “A woman in her 30s with prepaid insurance comes into your office today reporting that she has been increasingly depressed over the past two months, with disturbed sleep, decreased appetite, persistent hopelessness, and difficulty with concentration. She is not currently suicidal. She has missed a few days of work due to her depression this month. Her physical examination is normal.” We pretested this depression scenario during the development of the PIC instrument. The practicing clinicians in two focus groups of five or six clinicians found it realistic and thought that it contained the core information 73
needed for a decision about depression care management. In response to the scenario, clinicians rated their proclivity toward different management approaches as the first-line treatment: assess but not treat at this time, personally prescribe medications, personally counsel or provide psychotherapy, refer to mental health specialty, or refer to patient education or self-help program. Each of these five items was scored on a 5point Likert scale ranging from 1, very unlikely, to 5, very likely. We operationalized watchful waiting using the first item, assess but not treat at this time. For descriptive analyses, we created a dichotomous measure scored as positive if the clinician reported being likely or very likely to choose a watchful waiting approach and used the 5-point item in multivariate analyses. Primary care clinician characteristics Demographic and professional background. Clinician characteristics (from clinician report) included clinician type (internal medicine physician, general or family physician, or nonphysician, for example, physician assistant or nurse practitioner), board certification in their specialty, number of years in professional practice, gender, ethnicity, and the reported percentage of weekly visits with patients who have major depression (as a proxy for amount of experience treating depression). Knowledge. We included two separate scales to assess clinician knowledge of depression treatment, including a four-item psychotherapy scale and a six-item medication scale (details are available elsewhere) (12,14). Attitudes and reported practice. Clinicians reported whether they felt a “need to change or improve the way they manage patients with depression”; the item was scored as a binary measure to indicate “definitely” need to improve. We included two items to assess clinicians’ perceived skill about diagnosing depression and providing medication for depression. These items were scored from 1 to 4, with a score of 4 meaning “very skilled.” We also examined an item that asked clinicians about the percentage of pa74
tients (in a typical month) for whom they schedule a follow-up visit without starting treatment among patients they see with moderate to severe depression. Response options ranged from 1 to 5 (none, 25 percent or less, 26 to 50 percent, 51 to 75 percent, and 76 to 100 percent). System characteristics We included measures for selected barriers to optimal treatment for depression that have been identified in previous analyses (12,14) to be important predictors of depression-related practices. These barriers were “mental health professionals are not available,” “inadequate time to provide follow-up,” and “medical problems are more pressing.” Clinicians scored their perceptions of these barriers as 1, does not limit; 2, limits somewhat; or 3, limits a great deal. We included an indicator of whether or not the clinician reported using the guidelines of the Agency for Health Care Policy and Research (AHCPR, now the Agency for Healthcare Research and Quality) (23,24) for the treatment of depression, and we included a binary indicator for clinicians who routinely refer almost all patients with major depression to mental health specialty treatment. We also included a binary indicator of whether or not the clinician worked in a staff- or group-model organization compared with a network or independent practice model (14). Patient characteristics Depression severity. Patient characteristics were assessed with a set of rate measures representing the percentage of study patients with a given criterion for each clinician. We included an indicator of depression severity; the percentage meeting the cutoff point on a revised version (25) of the Center for Epidemiologic Studies Depression Scale (CES-D) (26). This revised scale is based on 13 items from the original 20-item CESD and ten new items that reflect changes made to the 4th edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). Possible scores on the scale range from 0 to 69, and the scale is rescaled to range from 0 to 100, first by taking PSYCHIATRIC SERVICES
the difference between the raw score and the minimum score in the patient sample and then by dividing through the range of the score among the sample patients. Chronic medical conditions, gender, and ethnicity. We also included rates of patients reporting two or more chronic medical conditions, rates of patients who were female, and rates of patients who were nonwhite. Because 12 of the 167 clinicians did not have visits with PIC study patients, the sample size was reduced to 155 in the multivariate analyses that included these rates. Statistical analysis We evaluated the face validity of the scenario-based clinician measure of watchful waiting by examining its association with clinicians’ reports of the extent to which they schedule a follow-up visit without starting treatment for their patients with moderate to severe depression. We used Student’s t tests and chi square analyses to examine bivariate relationships between each clinician, patient, and system characteristic and proclivity to choose watchful waiting for depression. Next, we examined the relative effects of the variables in multilevel regression models by entering the variables in sets according to our conceptual model, making sure to retain variables that were significant in bivariate analyses. In final models, we adjusted standard errors using the sandwich estimator (also known as robust variance estimator) (27) for the nesting of patients within clinic. We imputed missing data elements five times, forming five replicates of complete data. All analyses were conducted five times on each data set, and then results were synthesized by using Rubin’s multiple imputation inference procedure (28–30). We illustrated significant effects by calculating the adjusted means on the basis of parameters of the regression models.
Results Outcome validation Our scenario-based outcome measure of proclivity for watchful waiting was significantly correlated with clinicians’ reports of the proportion of
♦ ps.psychiatryonline.org ♦ January 2007 Vol. 58 No. 1
patients for whom they typically schedule a follow-up visit without starting treatment (r=.31, p
