Mental Health Diagnoses and Health Care Utilization in Persons With Dementia, Parkinson’s Disease, and Stroke

The higher prevalence of depression and anxiety associated with patients with neurological disorders, compared with the general population, is widely recognized, particularly in conditions affecting older adults.¹ Depression co-occurs in 30%−50% of individuals with Alzheimer’s type dementia and in 40%−50% of those with Parkinson’s disease (PD).^2,3 Estimates of poststroke depression range between 20% and 70%.⁴ Anxiety symptoms are estimated to affect 20%−50% of individuals with Alzheimer’s disease,⁵ PD,⁶ and stroke.⁷ Depression and anxiety hinder rehabilitation and are associated with increased pain, mortality, and impairments in activities of daily living.^5,8–10 Despite evidence for effective treatment options using pharmacotherapy, psychotherapy, and behavioral interventions for depression and anxiety associated with neurological disorders,^7,11,12 treatments are often not implemented. Barriers to specialty mental health care are numerous and include transportation, mental illness stigma, and limited access to mental health providers because of financial constraints, insurance coverage, or nonavailability of qualified clinicians.¹³

A unique window into mental health care utilization, in the absence of financial or insurance barriers to health care, is available through the analysis of veterans with neurological conditions who receive care through Veterans Administration health care services. Comparison of psychiatric diagnoses and mental health treatment across neurological conditions allows assessment of the relationship of diagnosis to health care patterns. Accordingly, this study used an administrative database to examine rates of anxiety and mood disorder diagnoses in older veterans across multiple neurological conditions (dementia, PD, and stroke), dementia only, PD only, or stroke only. Rates of mental health treatment (psychiatric medication use and specialty mental health outpatient visits) were examined among those veterans also diagnosed with depression or anxiety.

Methods

This study used administrative data from the Veterans Integrated Service Network 16 Data Warehouse, which includes demographic data, diagnostic codes for medical and psychiatric diagnoses (ICD-9-CM codes), outpatient and inpatient visits, and prescribed medications for 10 medical centers in the Veterans Administration Healthcare Network of the South Central region of the United States. The study was approved by the institutional review boards of Baylor College of Medicine and affiliated hospitals.

Subjects

Subjects identified from the database were men, age 55 years or older, with at least two Veterans Administration outpatient visits and, to increase the accuracy of diagnoses in the sample, with a diagnosis of PD, dementia, or stroke based on ICD-9-CM codes recorded at least twice in the medical record between October 1, 1997, and September 30, 2009.¹⁴ These 12 years were selected to provide a wide enough window to examine patterns of diagnosis and health care utilization that were available for research. Patients were not included in a diagnostic category if it was coded only once.

Neurological Diagnoses

Subjects were categorized as being in one of four diagnostic subgroups, based on their neurological diagnoses during the 12-year study period: (1) the multiple diagnoses group included subjects with more than one neurological condition, PD and stroke, PD and dementia, stroke and dementia, or all three (PD, stroke, and dementia); (2) dementia only; (3) PD only; and (4) stroke only. ICD-9 codes recorded for the neurological diagnoses included those for dementia (290.00–290.13, 290.20, 290.21, 290.30–290.43, 291.2X, 292.82, 294.10, 294.11, 294.8X, 331.0X, 331.11, 331.19, 331.82, 333.4X, 046.1X, 094.1X), PD (332.0), and stroke (430.XX-432.XX, 433.01, 433.11, 433.81, 433.91, 434.01, 434.11, 434.91, 436.XX, 437.1X, 438.XX).

Psychiatric Diagnoses

The data extraction included all ICD-9 anxiety and mood disorder diagnostic codes. Similar to the neurological diagnoses, the presence of depressive and anxiety disorders in a given subject in the analyzed dataset required two instances of the ICD-9 codes. However, the diagnosis of bipolar depression required (1) two or more instances of ICD-9 diagnosis of manic depressive or bipolar affective disorder, depressive, or (2) one instance of the ICD-9 code for manic depressive or bipolar affective disorder with current depression and at least one additional diagnosis at another time of another depressive subtype.

Primary Outcome: Mental Health Services Use and Medications

The dataset included antidepressant, antipsychotic, and mood-stabilizing medications. A patient was considered to have used a given medication if it was coded at least twice in any given year between October 1, 1997 and September 30, 2009.

Using clinic codes that distinguish where an appointment was located, we based rates of health care utilization on the presence and annual frequency of mental health outpatient visits. We considered patients to have attended a clinic if they visited at any point over the study period and calculated the frequency of visits as the average number of visits per year.

Analyses

Chi-square tests were used to investigate differences in presence of psychiatric disorders, medication use, and outpatient visits as a function of the neurological subgroup, i.e., multiple diagnoses, dementia only, PD only, or stroke only. The nonparametric Fisher’s exact test was used for associations with small cell sizes. We used one-way between-groups analyses of variance (ANOVAs) to examine neurological group differences in the means of patient characteristics and annual frequency of outpatient visits differed between neurologic categories. To correct for α inflation, tests of overall associations between neurologic categories and both categorical and continuous outcomes were evaluated at α≤0.01. Homogeneity of variance across the four neurologic groups was examined using the Brown Forsyth test, followed by Welch’s correction when the assumption was violated. Significant omnibus ANOVAs were followed up with post hoc pairwise comparisons, which were corrected for α inflation using Tukey’s wholly significant difference. Chi-square tests and ANOVAs were then used to conduct subgroup analyses to examine presence and annual frequency of mental health outpatient visits exclusively for those who had any anxiety or depressive diagnosis. To avoid deflating mean frequencies of outpatient visits due to mortality during the study period, mean frequencies of outpatient visits were calculated by taking an average of the number of visits any given person had during each year only for the years that he or she was alive. Analyses were performed with SAS version 9.3 (SAS Institute, Cary, NC).

Results

The dataset included 3043 male veterans diagnosed with dementia, PD, or stroke. Table 1 describes the demographic characteristics of the total sample and the neurological groups. In the multiple diagnoses group (N=690), 51 (7.4%) had PD and stroke; 80 (11.6%) had PD and dementia; 490 (71.0%) had stroke and dementia; and 69 (10.0%) had PD, stroke, and dementia. There were significant group differences in age across the neurological diagnoses [F(3,3039)=105.87, p<0.0001]. The stroke-only group was significantly younger than the groups with multiple diagnoses, dementia only, or PD only (Tukey-adjusted p<0.05). Race/ethnicity was related to type of neurological diagnosis [χ²(3)=19.57, p=0.002], such that a greater percentage of the PD-only group was Hispanic relative to the three other groups.

Prevalence of Depression and Anxiety Diagnoses

We initially examined prevalence of depression and anxiety among those groups of all dementia (dementia only and dementia with other neurological diagnoses, N=1190), all PD (PD only and PD with other neurological diagnoses, N=339), and all stroke (stroke only and stroke with other neurological diagnoses, N=2273). Among everyone in the total sample with a diagnosis of dementia, 351 (29.5%) had a depressive diagnosis; 186 (15.6%) had any anxiety diagnosis; and 61 (5.1%) had bipolar diagnosis. Among those with PD, 75 (22.1%) had a depressive, 44 (13.0%) had an anxiety, and 21 (6.2%) had a bipolar diagnosis. Among those with any diagnosis of stroke, 456 (20.06%) had a depressive disorder, 246 (10.82%) had any anxiety diagnosis, and 41 (1.80%) had a bipolar diagnosis.

Table 2 reports rates of depression and anxiety in the total sample and across the exclusive neurological groups: dementia only, PD only, stroke only, and multiple diagnoses. Over the 12-year study period, 25.8% had both depressive and anxiety diagnoses. The neurological diagnosis was related to the presence of a depressive disorder [χ²(3)=75.39, p<0.0001]. Relative to the PD-only and stroke-only groups, subjects in the multiple diagnoses and dementia-only groups were more likely to have a depressive or bipolar disorder diagnosis. Rates of depression and bipolar diagnoses were comparable between the multiple diagnoses and dementia-only groups. In this dataset, the PD-only group had the lowest rates of any depressive or bipolar disorder diagnoses.

Subjects in the multiple diagnosis group, followed by the dementia-only group, were more likely to have an anxiety disorder diagnosis. The most prevalent anxiety disorder diagnoses were posttraumatic stress disorder (Fisher’s exact, p=0.001), panic disorder with or without agoraphobia (Fisher’s exact, p=0.005), and anxiety disorder not otherwise specified [χ²(3)=16.14, p=0.001]. Compared with all the single diagnosis groups, subjects in the multiple diagnoses group had higher rates of posttraumatic stress disorder and anxiety disorder not otherwise specified diagnoses. Relative to the PD-only and stroke-only groups, the multiple diagnoses group also had higher rates of panic disorder diagnoses. The dementia-only group had a higher rate of any anxiety disorder than the PD-only and stroke-only groups.

Health Services Use

Among the patients with any depressive and/or anxiety diagnosis (N=786, 25.8% of the total sample), 369 (46.9%) had at least one mental health outpatient visit at any point over the study period. Within this subset of subjects with diagnosed anxiety or depression, mental health clinic visits occurred in a higher proportion of the multiple diagnoses and dementia-only subgroups (49.9% and 40.4%, respectively), relative to the PD-only and stroke-only subgroups (1.6% and 8.1%, respectively; p<0.0001 for all comparisons). Additionally, within the subset of subjects with anxiety or depression diagnoses, the dementia-only subjects had more mental health visits than those with stroke only [mean (standard deviation)=2.42 (4.20) versus 0.33 (0.40), respectively; p=0.004).

Medication Use

For the subset with diagnosed anxiety or depression, 67.1% with multiple diagnoses were prescribed an antidepressant, 28.1% an antipsychotic, and 24.5% a mood stabilizer (Table 3). The stroke-only subjects with anxiety or depression had the highest rates of prescriptions for antidepressants (74.5%) or mood stabilizers (31.6%) compared with the other neurological conditions.

Discussion

It is widely recognized that depression and anxiety disorders frequently complicate the care of patients with dementia, PD, and stroke.^4,5,15,16 This investigation of an administrative database of older male veterans with diagnoses of stroke, PD, or dementia revealed that individuals with more than one of these neurological diagnoses and those with dementia only have even higher rates of mood and anxiety disorder diagnoses than those with stroke or PD only. Specialty mental health outpatient care was provided to almost half of those with dementia only and a mood or anxiety diagnosis, and about two thirds of all subjects with mood or anxiety disorders were prescribed antidepressants.

These data further suggest discrepancies in the clinical recognition of depressive and anxiety disorders in patients with neurological disorders. Whereas our rates of depressive diagnoses across the neurological conditions were similar to prevalence estimates in other studies,^3,4,17 we observed a lower prevalence of anxiety diagnoses in our sample than in other epidemiological studies on these populations.^5,6,18 Screening for depression is recommended in primary care and for those with the neurological conditions in this study.^19–21 However, there has been less emphasis on anxiety screening in routine primary care or neurology clinic settings. Given that both anxiety and depressive disorders cause added distress and dysfunction to the neurological condition^5,18,22 but involve different treatment approaches, recognition of either condition should be an important part of routine clinical care.

Antidepressants were prescribed for the majority of patients with depression and/or anxiety, but most subjects with PD only or stroke only did not receive specialty mental health care. This lack of mental health collaboration for PD and stroke is consistent with the literature^23,24 and highlights an opportunity to improve health care for this population. Integrated mental health care, in which patients receive mental health interventions in non–metal health clinical settings, could reduce some barriers to receiving treatment for depression and anxiety and enhance clinical outcomes.²⁵

Across all neurological disorders, antidepressants were the most common type of treatment. The acute nature of stroke, the robust literature on the negative impact of poststroke depression on recovery,⁸ and the importance of engagement in rehabilitation after a stroke may contribute to the higher rates of antidepressant treatment in the stroke-only subgroup versus those with other neurological conditions. The adverse health and safety side effects of antipsychotic medications in dementia patients have led to declining use.²⁶ However, in our sample, antipsychotic prescription rates were highest in the subset of patients with dementia only plus an anxiety or depressive disorder. Treatments to manage mood and anxiety are of interest for further study and implementation, given the increased risk of poor physical outcomes, such as diabetes, and mortality^27,28 in patients with dementia on antipsychotics.

Several limitations of the study warrant recognition and constrain generalizations on the prevalence of psychiatric diagnoses and health care practices for patients with neurological diagnoses. In particular, the sample was restricted to veterans receiving treatment through the Veterans Health Administration and, as such, subjects should not face insurance-related barriers to receiving specialty mental health care. In addition, diagnoses were based on medical records only, and rates of unidentified depressive or anxiety disorders, the effectiveness of psychiatric medications, or recommendations of clinical psychiatric treatments that were not followed cannot be discerned from the database.

This analysis of a Veterans Administration database underscores the need to further characterize the complex health care needs of patients with neuropsychiatric disturbances and develop strategies that maximize clinical outcomes for both veterans and those in the general population. This study also adds to the literature on recognition and health care use for veterans with dementia, PD, and stroke and co-occurring anxiety and depression. Treatments such as antidepressants are frequently used to address depression and anxiety in dementia, PD, and stroke; however, care remains unstandardized. Clinical outcomes will benefit from routine mental health screening with follow-up involving integrated care teams.²⁹