Participants
This study was conducted as part of a larger investigation examining depression and anxiety in HF patients. Potential participants were identified from a large Veterans Affairs (VA) hospital through the Outpatient Care File and Patient Treatment File databases of the Veterans Health Administration (October 2004 through September 2006) using International Classification of Diseases, Ninth Edition, Clinician Modification (ICD-9-CM) codes (398.91, 402.01, 402.11, 402.91, 404.01, 404.03, 404.11, 404.13, 404.91, 404.93, 428, 428.0, 428.10, 428.9). Inclusion criteria were determined through several stages. First, medical records of veterans 60 years or older who had a documented diagnosis of HF (N = 4129) were extracted through a database search. Review of medical records by a research assistant led to the exclusion of veterans who did not have contact information, lived outside the study site area, were deceased, or did not have sufficient medical record documentation of HF (N = 2834). The remaining 1295 participants received a letter via postal service indicating that a study coordinator would contact them by phone unless they opted out of the study. Of those mailed a letter, 443 patients were not interested in participating, 361 were unable to be contacted by phone, and 187 were deceased or too frail to participate. For the larger study, 304 telephone screens were conducted. The final participant sample for this study (N = 158) was enrolled between November 2006 and April 2007 and included all consecutive participants who underwent a telephone screen and completed all study measures during the first 6 months of the larger study. This project was approved by the Houston VA Research & Development Committee at the Michael E. DeBakey VA Medical Center and the Institutional Review Board at Baylor College of Medicine, and informed consent was obtained from all patients.

Procedures
Telephone assessment.
The telephone assessment measures included (1) a brief demographic survey; (2) a semistructured interview to determine New York Heart Association (NYHA) criteria (only patients with symptoms of class II or greater were included for further study); and (3) a short, 6-item cognitive screening instrument to exclude participants with significant cognitive impairment (participants with 3 or more errors were excluded). All participants remaining eligible were administered the Geriatric Depression Scale (GDS) and the Geriatric Anxiety Inventory (GAI).

Medical-record review.
Each patient’s EMR was reviewed by a trained research assistant using a structured extraction form. Reviews of EMRs were conducted for the 12 months of care provided prior to each patient’s telephone screening, as well as the 6-month period of care following the assessment. This before-and-after medical-record-review approach was deemed necessary to accurately assess the potential of the health care system to detect the presence of depressive or anxiety symptoms that may have been long-standing or recent in onset.

Data extracted from patient medical records included number of outpatient encounters (separately assessed for 12 months prestudy and 6 months poststudy assessments), screening information related to depression (conducted by the health care system via routine primary care/preventive medicine screenings using a 2-item instrument), diagnosis of depression or anxiety (outpatient or inpatient) in the medical record, documentation of antidepressant or antianxiety medications, and presence or absence of mental health clinic visit(s).

Structured reviews of the medical records were completed by 2 raters, using a structured format to increase reliability of ratings. A random 10% of medical records were reviewed by a third rater and compared for accuracy with results of the primary reviewers. Raters had a 95% agreement and a ê coefficient of 0.90.

Measures and Study Variables
Geriatric Depression Scale.
The 15-item GDS was used to determine participant self-reported symptoms of depression. The GDS is a widely used, reliable, and valid measure for assessing symptoms of depression among older adult populations. This measure is particularly useful with medically ill patient populations because of its lack of inclusion of somatic-based symptoms of depression (for review, see Stiles and McGarrahan) and has been used specifically in HF populations. The present study administered the GDS via telephone based on prior work showing reliability and validity of the GDS by telephone. As suggested by the original authors, a traditional cutoff of 6 or greater was used to determine clinically significant symptoms of depression.