Geriatric Anxiety Inventory.
The GAI is a self-report measure of anxiety with 20 items to which the participant answers yes or no for how he or she has felt for the last week. The GAI was also administered via telephone. It was developed to assess the symptoms of anxiety specifically among older adults. The GAI has demonstrated good reliability and convergent validity but lower divergent validity, as it correlates higher with measures of depression than with other measures of anxiety. For the current study, a cutoff score of 9 or greater, as suggested by the original authors, was used to determine the presence of clinically significant anxiety.

Medical-record review.
Recognition of depression or anxiety. For the purposes of this study, recognition of depression or anxiety was defined as documentation in the medical record of any of the following:(1) an outpatient diagnosis of depression or anxiety, (2) an inpatient diagnosis of depression or anxiety, or (3) any documentation or references to anxiety or depression within the outpatient progress notes.

Provision of mental health treatment. Mental health treatment was considered to have been provided if 1 or both of the following were present in the medical record:(1) any mental health encounters/clinic visits (visits included those by any mental health practitioner, including psychiatrists, psychologists, nurses, social workers, and physician assistants) and (2) documentation of a psychotropic medication for depression or anxiety. Medications included all antidepressant and anxiety medication classes except trazodone and amitriptyline. The medical-record documentation of these medications was examined in detail to ensure that the medication(s) were prescribed for depression or anxiety rather than for another clinical condition, such as sleep or pain. If the medications were documented as prescribed for another health issue, treatment was not noted. If the medications were not documented, the default was to consider the presence of these medications as evidence of mental health treatment.

Notably, the concept of mental health treatment, for the purposes of this study, did not assume recognition. Although it is conceivable that practitioners who prescribe a psychotropic medication likely recognized a mental health condition, it was believed that the concept of recognition was best described as a specific action taken by a provider to notate the medical record and, potentially, to alert other practitioners of the identified condition or symptoms.

Other variables.
Illness burden (relative-risk score). The illness burden, or relative-risk score, was assessed using a diagnosis-based, risk-adjustment methodology (DxCG Company, Boston, Mass.) validated in the VA population. An individual’s relative-risk score is calculated using database information to identify his or her total predicted health care costs, which are then divided by the average predicted cost of the population. A score of 1.0 reflects an average risk; scores below 1.0 represent less illness burden, and scores above 1.0 reflect increased burden.

NYHA functional classification. The NYHA functional classification was used to assess the degree to which heart failure limits physical activities corresponding to disease severity. As functional abilities decrease, NYHA classification increases from class I to class IV. In the current study, NYHA classification was determined using a semistructured clinical interview completed by trained research assistants.

Analyses
Descriptive statistics were used to provide information on the prevalence of depression and anxiety symptoms in HF patients and to describe the percentage of patients with mental health recognition and treatment, as determined by EMR review. Independent sample t tests and x² tests were used to examine differences between HF patients classified as depressed or anxious versus those not depressed and not anxious, using the GDS and GAI clinical cutoff scores. Classification tables were used to examine the percentage of patients identified by self-report versus that recognized in the EMR, as well as those with a documented recognition of depression or anxiety versus those receiving documented depression or anxiety treatment.

Logistic regression procedures were used to determine patient-level factors associated with recognition of depression or anxiety in the medical record. Predictors included demographic variables (age, ethnicity), health-service-use variables (categorical number of outpatient encounters pre- and post-assessment), and clinical indicators (GDS total score, GAI total score, NYHA classification, and relative-risk score). A second logistic regression was used to explore predictors of notation in the medical record of treatment for depression and/or anxiety, using the same variables as above, with the addition of a dichotomous variable representing the presence or absence of a positive primary care depression screen.