Dr. Kirkcaldy is a hospitalist in the Departments of Internal Medicine and Psychiatry, Montgomery General Hospital, Olney, Maryland, and a student at the Johns Hopkins Bloomberg School of Public Health, Baltimore. Dr. Tynes is medical director, Jefferson Parish Human Services Authority, New Orleans, and associate professor of clinical psychiatry, Department of Psychiatry, Tulane University School of Medicine, New Orleans. Send correspondence to Dr. Tynes at Departments of Psychiatry and Internal Medicine, Tulane University School of Medicine, 1440 Canal Street TB53, New Orleans, LA 70112 (e-mail: email@example.com). William M. Glazer, M.D., is editor of this column.
Despite a high prevalence of depression in the primary care setting (1,2) and significant associated disability and health care costs (3,4), depression is unrecognized in one-third to one-half of affected primary care patients (1,2,3,4,5). Recognizing depression screening as a best practice, the United States Preventive Services Task Force issued recommendations for regular depression screening of adults (6), and in 1998 the U.S. Department of Veterans Affairs (VA) mandated annual depression screening at VA primary care clinics nationally. The VA Medical Center in New Orleans implemented annual screening with a four-item questionnaire based on a screening tool validated by Rost and colleagues (7) (see box on the next page).
Depression screening tool used at the Veterans Affairs Medical Center, New Orleans
An affirmative response to any of the questions represents a positive screen.
1. In the past year, have you had two consecutive weeks or more during which you felt sad, blue, or depressed or when you lost all interest or pleasure in things that you usually cared about or enjoyed?
2. Have you had two or more years in your life when you felt depressed or sad most days even if you felt okay sometimes?
3. Have you felt depressed or sad much of the time in the past year?
4. In the past week, did you feel depressed for one day or more?
Screening at this site is performed before the clinical encounter with the physician or nurse practitioner. An intake nurse, who also documents the vital signs, performs the screening in a private area and documents the responses in a text box at the bottom of the electronic medical record screen. In some cases the responses are documented on a nursing note that is included as a progress note. The results can be as simple as "Screen is positive." In some cases more thorough notes identify specific items on which the individual screened positive. A text box in the electronic medical record highlights when the annual screening is due, which acts as a prompt for the intake nurse to perform the screening.
Using a retrospective cross-sectional study design, we examined the implementation of the screening program at the VA Medical Center in New Orleans. The goals of the investigation were to establish a benchmark for the current level of screening and management and to identify barriers to screening that may have emerged during the implementation phase. A waiver of institutional review board approval was obtained because this was a quality improvement initiative.
The Primary Care in Internal Medicine (PRIME) clinic at the VA Medical Center, New Orleans, was the study site. The clinic structure is organized into three firms to which providers and patients are randomly assigned. The providers include nurse practitioners, internal medicine house-staff, and internal medicine attending physicians. The eligible study sample consisted of all adults attending the PRIME clinic for a provider visit during the month of January 2002. Because of time and resource limitations, all charts could not be reviewed. Alphabetically by team and provider, starting with the first provider on one team, data for patients were recorded until 1,100 patients were included. This encompassed two of the three teams of providers. The time period reviewed in each record was January 2001 to April 2002.
Several data sources were reviewed: the Computerized Patient Record System electronic medical record, pharmacy records, referral records, encounter forms, nursing intake notes, and outpatient and inpatient clinician notes. A patient was considered screened if screening results were documented within the past 12 months and if screening occurred on a visit to a physician or a nurse practitioner. The screening was considered positive if the encounter form or intake nursing note documented "Screen positive" or "Pt. referred to provider" or if affirmative responses to screening questions were noted. Ambiguous screening results were excluded from the analysis. Clinical management was assessed by using clinic notes, pharmacy records, and psychiatric referral records. Treatment or follow-up was deemed adequate if a prescription for an antidepressant was offered or a referral was made for mental health services, based on current treatment guidelines of the American Psychiatric Association (8). The rates of screening and management of positive screens were calculated.
Of 1,100 patients, all but two were men. The mean±SD age of the sample was 62.9±12.3. A total of 461 patients (42 percent) were white, 455 (41 percent) were black, eight (.7 percent) were Latino, and one (.1 percent) was Native American. Race or ethnicity was unlisted for 174 patients (16 percent).
Universal screening was nearly achieved. A total of 1,068 patients (97 percent) were screened, either at the January 2002 visit or at some time in the previous 12 months. Eighty-one of the patients who were screened (8 percent) reported that they were currently receiving treatment for depression. Seventy of the 1,068 patients (7 percent) screened positive, and 50 of these patients (71 percent) were offered at least one intervention. The interventions offered included psychiatric referral (17 patients, or 34 percent), continuation of an antidepressant or mood stabilizer (26 patients, or 52 percent), a new prescription for an antidepressant or mood stabilizer (12 patients, or 24 percent), or a combination of strategies. Four patients refused treatment. It is unclear how many patients declined screening. Only 32 of the 1,100 patients (3 percent) were not screened, but the reasons for their lack of participation were not documented.
The results of this evaluation establish a benchmark value for the frequency of screening and for the basic management of positive screens. We found that regular screening was performed successfully, as evidenced by the excellent screening coverage (97 percent). The 7 percent rate for positive screens is comparable to previously published data on detection rates for primary care depression screening (5,9). On the basis of the high rate of screening, we conclude that depression screening is acceptable to this population. The frequency of follow-through on the positive screens, however, was not as robust. Nearly one-third of patients who screened positive did not receive a basic level of depression management, which highlights a limitation of the screening program that invites further improvement.
Nearly universal screening and a 70 percent frequency of management of positive screens does indicate a relatively successful implementation. A high rate of detection may be fostered by several factors. A short, user-friendly, and validated screening instrument is critical. Visible reminders to perform the screening, access to the screening instrument at the point of care, and a simple manner of recording results—with each of these elements incorporated into the medical record—are likely to enhance the rate of screening. Performance of the screen by the intake nurse seems to be acceptable to patients and an efficient use of resources. The manner of communicating the results to the physician or physician-extender should be efficient and informative and should promote ease of use. In this program, the results of depression screening are placed in the same window of the medical record as the other primary care screening results.
Several possible barriers to effective follow-up were identified during our investigation. The utility of the screening tool may have been limited by incorrect administration of the instrument, which resulted in documentation of contradictory results. There was documentation of negative results despite affirmative responses to the screening questions, as well as individual screening results that were scored as both positive and negative. Incorrect administration of the instrument occurred in only a small number of cases, however. As noted above, an important barrier to follow-up is related to how the screening results are displayed. Results that are not prominently documented can be easily overlooked by busy clinicians.
Other barriers to depression screening and management have been proposed by Preboth (10). Patient-related factors, such as stigma, minimization of distress, and reluctance to disclose symptoms, will limit detection. Physician-related factors, such as time-pressures, closed-ended interviewing styles, inadequate knowledge of the disease process, and discomfort discussing depression, have been proposed. These factors may account for the low rate of depression detection by house staff (11,12). Again, the mode of delivery of positive screening results can have an impact on the translation of screening into clinical action. Face-to-face contact between the screening nurse and the physician and the provision of a depression treatment protocol for positive screens appears to improve detection by the physician and clinical management (13).
Trials of depression screening that have fared well have employed integrated screening programs that include educational materials, telephone follow-up and other services; feedback and treatment advice to providers or education of physicians and patients; and access to mental health services (9,14). Finally, the electronic medical record currently in use at the VA clinic does not have a clear format to ensure management of patients with positive screens. The system relies on the clinician to notice the screening results. Incorporating "forced" acknowledgement of the screening results at the point of care may increase the frequency of follow-up.
The relatively high rate of depression screening demonstrated by our data suggests that depression screening in primary care is a feasible objective. Depression screening may not be feasible in all clinical settings, especially those that do not have access to mental health services and psychiatric consultations.
Increasing the rate of management of positive screens and follow-up will depend on further improvement in the VA clinic's screening program. Patient-related improvements might enhance patient education about depression and treatment options. Possible physician-related improvements include education to further improve competence and comfort with screening for and managing depression.
System-related improvements are likely to be the most effective and might include more prominent display of the screening results with "forced" acknowledgment of the positive screens. The system could be further enhanced by the addition of "point-and-click" tools, such as diagnostic criteria, a differential diagnosis, management advice, and treatment algorithms, along with "one-click" medication ordering that is coded with "standard of care" dosing, recommended follow-up, and dosage titration. Additionally, a small change in the screening system that requires documentation of the responses to the screening questions through "forced choices" may enhance the utility of the results for both clinicians and researchers. When a screen is positive, face-to-face contact between the screening nurse and the physician may also improve detection by the physician.
Limitations of this study include the gender homogeneity and older age of the sample, which impair the ability to generalize the results. Patients from only two of three firm teams were included because of time and resource limitations, which may have introduced a selection bias. Because patients and providers are randomly assigned to the teams, we believe that the risk of selection bias is small. Finally, reliance upon positive screens in lieu of psychiatric interviews did not allow us to distinguish between positive screens and a diagnosis of major depression based on DSM-IV criteria. Reliance on screens increases the risk of false-positive results.
Despite these limitations, we believe that these data represent accurate approximations of benchmark values for depression screening.
Primary care clinics and offices are encouraged to pursue implementation of routine depression screening. We have highlighted a number of ways that such a program can be successfully implemented, as well as possible barriers to implementation that can be overcome. Further research and quality improvement initiatives are needed to move closer to universal screening as a best practice for the management of depression.