Clinical Care Quality Among Veterans Health Administration Patients With Mental Illness Following Medical Home Implementation

The burden of medical conditions is higher among individuals diagnosed as having mental illness compared with those without mental illness (1, 2). Individuals with schizophrenia, bipolar disorder, or major depressive disorder are more likely to experience obesity and diabetes, cardiovascular diseases, and musculoskeletal diseases, among others (3). Individuals with posttraumatic stress disorder (PTSD) are at increased risk of cardiovascular disease and diabetes (4). Substance use and substance use disorders have also been associated with increased risk of a number of negative health outcomes (1, 5). The co-occurrence of mental illness and chronic medical conditions has been linked with greater functional and occupational impairment, reduced quality of life, and higher health care costs compared with individuals with only one of these conditions (1, 2, 6–8). Many studies have shown that individuals diagnosed as having mental illness die prematurely (9, 10), often due to preventable causes, such as diabetes and cardiovascular disease (11).

Delivery of poorer-quality preventive and chronic disease management care may play a key role in contributing to poor general health outcomes among those with mental illness (12, 13). Such services may be particularly important among U.S. veterans seen within the Veterans Health Administration (VHA), a population that experiences disproportionately high rates of mental and medical illnesses (14–16). Past studies conducted within the VHA showed that veterans with mental illness were less likely to receive certain forms of preventive care (e.g., cancer screening) and chronic disease management care (e.g., diabetes performance measures) compared with those without mental illnesses (17–20).

Previous research has focused largely on quality indicators related to a single preventive procedure (e.g., cancer screening) or medical condition (e.g., diabetes) or have examined outcomes in veterans with or without “any mental illness” or veterans with a select few mental illnesses (17–20). Since these initial studies, the VHA has undertaken two large-scale national implementation efforts to enhance the primary care services offered in VHA facilities. In 2007, the VHA began to systematically integrate mental health care services into the primary care setting (primary care–mental health integration [PC-MHI]; 21). In 2010, the VHA also became the largest integrated health care system to implement a patient-centered medical home model (22). The VHA model, called Patient-Aligned Care Team (PACT), was designed to deliver coordinated, patient-centered care by improving access and providing care management through continuous team-based care. Core PACT teams consist of a primary care provider, a nurse care manager, a clinical associate (licensed practical nurse [LPN] or LPN equivalent), and a clerk (23).

Importantly, the PACT model was designed to include both behavioral health specialists as well as PC-MHI providers within the expanded PACT teams (23). As a result, the PACT model may be well positioned to provide high-quality care to veterans with mental illness. This study sought to evaluate the quality of preventive care and chronic disease management care among VHA primary care–enrolled patients with and without mental illness following the nationwide implementation of PACT.

Methods

The study used an observational cohort design that included veterans seen within VHA primary care between 2010 and 2013. This study was deemed exempt from institutional review board approval because it was nonresearch and analyzed nonidentifiable data. The Primary Care Management Module contained within the VHA’s Corporate Data Warehouse was used to identify all veterans who were enrolled and assigned to a VHA primary care provider in the 3 years after PACT implementation. An initial observational cohort of 4,461,247 veterans who had an assigned primary care provider and who attended at least one primary care appointment within the VHA between April 2010 and March 2011 was identified.

Because many of the quality indicators of interest are completed on an annual basis, this identification method was then used again for the following periods: 2011–2012 and 2012–2013. Additional information regarding methods used to define the primary care cohort are detailed elsewhere (10). Patients were then categorized into six mental illness groups: depression, PTSD, anxiety disorder, serious mental illness (bipolar or schizophrenia), substance use disorder, any mental illness (any of the five diagnoses) or a comparison group with no mental illness (none of the five diagnoses) (10).

Because veterans often have more than a single psychiatric condition, the mental illness groupings were not mutually exclusive (e.g., a veteran with PTSD and depression was included in both the PTSD and depression groups). A substance use disorder was considered present if there was one or more hospitalization or outpatient visit with a relevant ICD-9 diagnosis code. We considered all other mental illness conditions to be present if there were two or more outpatient visits or one or more inpatient hospitalizations with a relevant ICD-9 diagnosis code (10).

Although veterans with some mental illnesses (e.g., eating disorders) may have been included in the no mental illness comparison group, the prevalence of these disorders (particularly without comorbid depression, PTSD, anxiety disorders, serious mental illness, or substance use disorders) appears to be low in the VHA (24).

Analytic Sample

Data from the VHA External Peer Review Program (EPRP) were used to examine quality of care. The EPRP is an external audit program that assesses clinical performance by using standard criteria based on clinical care guidelines (25). EPRP data are abstracted manually from the electronic health record by an independent external contractor and reported by fiscal year (October–September). Interrater reliability (k=0.9) measured among abstractors in the EPRP program has been high (26). The analytic sample for this study consisted of primary care patients whose electronic medical records were selected for review as part of the EPRP. Approximately 5% of veterans who were enrolled and assigned to a VA primary care provider were selected for EPRP review in each of the 3 years examined (5.3% in 2010, 4.6% in 2011, and 4.8% in 2012). Thus the sample size for the analytic sample ranged from 210,864 to 236,421. Denominators for each quality metric varied based on the specific inclusion-exclusion criteria for that indicator.

Quality-of-Care Indicators

Thirty-three quality indicators assessed by the VHA EPRP were examined. Quality indicators related to preventive care included tobacco screening, cancer screening, and immunizations. Quality indicators related to the management of common chronic diseases included diabetes mellitus, hypertension, ischemic heart disease, and chronic heart failure.

Patient Characteristics

To characterize the sample, additional descriptive information about patient demographic characteristics (age, gender, race-ethnicity, and marital status) and prevalence of mental illness and common medical conditions (diabetes, hypertension, ischemic heart disease, and chronic heart failure) was abstracted from the VHA’s Corporate Data Warehouse for the initial cohort of veterans identified (veterans enrolled and assigned to a VHA primary care provider from April 2010 through March 2011; N=4,461,247). Because the sample size varied across the 33 clinical quality indicators in each of the 3 years examined, similar data are not presented for each analytic sample. [For ICD-9 codes used to define mental illness and medical conditions, see online supplement.]

Analyses

Our goal was to describe the proportions of eligible patients who met the 33 preventive care and chronic disease management clinical quality indicators across the six mental illness groups and the group without mental illness. The proportion of patients meeting a quality indicator in each mental illness group was compared with the proportion of patients without mental illness meeting the same quality indicator. These descriptive analyses were conducted to examine the quality of overall care provided within PACT irrespective of clinical site or patient characteristics. In our preliminary descriptive analyses, we determined that most quality indicators were met for at least 80% of all veterans with and without mental illness, and were frequently over 90%. Such a ceiling effect could obfuscate disparities between groups. Moreover, because our large sample size would make Type I error likely and result in statistically significant differences that were not clinically meaningful, we focused our efforts on identifying minimum clinically important differences when comparing adherence to quality indicators.

Because there are no current guidelines on detecting such differences when comparing adherence to quality indicators, we consulted with VHA primary care and mental health subject matter experts to determine a meaningful metric to compare veterans with and without mental illness. Through consensus, we determined that a consistent difference of at least 5% across all years examined would be meaningful in evaluating VHA policy. Therefore, we compared the proportion of veterans with and without a mental illness to identify the quality indicators with differences of at least 5% in 2010–2011, 2011–2012, and 2012–2013.

Results

Patient characteristics are presented in Table 1. The proportion of eligible veterans meeting each quality indicator within the depression, PTSD, anxiety disorder, serious mental illness, and substance use disorder mental illness groups can be found in Table 2. (For number of patients eligible for each indicator and graphs depicting quality indicator completion for veterans with and without any mental illness, see the online supplement.)

Approximately 25% of the initial cohort were diagnosed as having at least one of the five included mental illnesses, with 13.2% of veterans diagnosed as having depression, 9.1% with PTSD, 4.7% with an anxiety disorder, 8.1% with a substance use disorder, and 3.2% with a serious mental illness. Across the three time periods examined, the quality indicator met by the highest proportion of veterans was tobacco screening; nearly all veterans received this preventive care (>99.1% ). Conversely, the quality indicator met by the lowest proportion of veterans was influenza immunizations specifically among veterans without mental illness (65.2%, 64.7% , and 64.9%; Table 2).

For most (27 of 33) of the clinical quality indicators, there was no consistent clinically important difference of 5% or greater between those with and without mental illnesses. For six indicators, there were consistent differences of ≥5% between those with and without mental illness in each of the three time periods examined, with differences ranging from 5.0% to 12.0% (Table 3). A greater proportion of veterans (ages 50–64) received an influenza immunization in five of the six mental illness groups (any mental illness, depression, PTSD, anxiety disorder, and serious mental illness) compared with veterans without mental illness. Additionally, a greater proportion of veterans with chronic heart failure had documented left ventricular ejection fraction (LVEF) in three of the six mental illness groups (any mental illness, anxiety disorder, and substance use disorder) compared with veterans without mental illness.

Conversely, the proportion of veterans meeting the preventive care quality indicator for colorectal cancer screening was lower for veterans with a substance use disorder or a serious mental illness than among veterans without mental illness. Likewise, the proportion of veterans meeting the chronic disease management indicator for having acetylsalicylic acid (ASA) included in their current medications was lower for veterans with a substance use disorder and diabetes. The proportion of veterans meeting the indicator for having angiotensin-converting enzyme (ACE) inhibitor or angiotensin-receptor blocker (ARB) therapy included in their current medications was also lower for veterans with a substance use disorder and diabetes. Similarly, fewer veterans with a serious mental illness and diabetes met the same indicators related to having ASA and/or ACE inhibitor or ARB therapy in their current medications. Additionally, fewer veterans with a serious mental illness and ischemic heart disease met indicators for use of ACE inhibitor or ARB therapy.

Discussion

We found that veterans with mental illness generally received high-quality preventive and chronic disease management care, similar to those without mental illnesses, in the 3 years following the national PACT implementation, the VHA’s patient-centered medical home model. Consistent clinically important differences in the proportion of veterans meeting quality indicators were detected in just six of 33 indicators; furthermore, in two of these cases, the proportion achieving the quality indicator was in fact higher among those with mental illnesses. Contrary to earlier reports of lower quality care for those with mental illnesses before PACT implementation, we found few disparities between veterans with and without mental illness. Such findings are consistent with previous investigations demonstrating high-quality care in the VHA compared with similar metrics from the private sector (27, 28).

Influenza immunizations for veterans ages 50–64 were higher among individuals with any mental illness, depression, PTSD, anxiety disorders, or serious mental illness compared with veterans without mental illness. This finding is contrary to earlier VHA-based research showing that influenza immunization rates were higher among veterans without mental illness compared with those with psychiatric conditions or substance use disorders who were at least age 65 or considered high-risk (18). Increased receipt of influenza immunization may have resulted from overall increased VHA health care utilization by veterans with mental illness, including the VHA’s provision of access points outside primary care that do not require an appointment (8).

We found higher proportions of documented LVEF among veterans with chronic heart failure and any mental illness, anxiety disorders, or substance use disorders compared with veterans with no mental illness. As previously noted, increased health care utilization is often observed among individuals with mental illness, possibly providing more opportunities for screening and care (8). It is also possible that knowledge of health risks associated with long-term illicit drug use, including negative cardiovascular effects (5), contributes to providers being more likely to monitor heart functioning in individuals with substance use disorders. Additionally, anxiety disorders often include somatic symptoms (e.g., chest pain or discomfort experienced during panic attacks) (29, 30),which may prompt primary care providers to monitor or rule out other health conditions.

The proportion of veterans meeting a quality indicator was lower among veterans with mental illness for one prevention indicator (colorectal cancer screening) and three chronic disease management indicators (ASA included in current medication for individuals with diabetes, ACE inhibitor or ARB therapy included in current medication for individuals with diabetes, ACE inhibitor or ARB therapy included in current medication for individuals with ischemic heart disease with LVEF <40%). However, these differences were only detected when comparing veterans with substance use disorders or serious mental illness with those with none of the five included mental illnesses.

Such findings align with prior studies suggesting failure to meet care quality indicators may be more common among veterans with substance use disorders or serious mental illness (17, 19). Additionally, a prior study of veterans enrolled in VHA primary care found that those diagnosed as having a substance use disorder, serious mental illness, or depression demonstrated poorer outcomes (e.g., were more likely to be hospitalized or die within 1 year when compared with veterans without that mental illness; 10). Although widespread disparities in care quality were not observed, it remains problematic that these differences were detected, particularly given the health risks already faced by individuals with a substance use disorder or serious mental illness (3, 5). It is notable that the proportions of veterans receiving colorectal cancer screening were higher in this study than in prior VHA-based work, suggesting promising overall improvements in VHA care delivery (18).

The PACT model was designed to improve access to and coordination of care and may represent an important opportunity to reach veterans with mental illness. However, the PACT model was not designed to provide the longer-term mental health care that is often indicated for individuals with substance use disorders or serious mental illness. Patients with more severe levels of acuity, whose mental health care needs are unlikely to be met by the brief interventions offered through PC-MHI, are often referred to specialty mental health care (31). Furthermore, research has suggested that veterans with substance use disorders or serious mental illness may be less likely to engage in primary care services (32, 33). Thus veterans with a substance use disorder and/or a serious mental illness may have less contact with their PACT and PC-MHI providers and may have less exposure to the coordinated care provided in this setting compared with veterans with other forms of mental illness (e.g., depression or anxiety). Taken together with the results of this study, such findings suggest that engagement in care and certain aspects of care delivered (e.g., colorectal screens) may need to be enhanced for these high-risk groups.

Improving the understanding of the role that individual-level (e.g., beliefs about and trust in medical care, demographic characteristics, clinical differences) and system-level (e.g., PACT model implementation status, use of integrated behavioral health care) variables play in the delivery and receipt of high-quality care may allow providers to further leverage the PACT model and enhance care provided to veterans with and without mental illness. Additional research is also needed into the complex factors that contribute to poor health among individuals with mental illness (e.g., impact of chronic stress on the body; poor health behaviors such as tobacco and alcohol use, physical inactivity, and poor diet; side effects of psychiatric medications).

Understanding these variables may provide insight into how best to provide care for veterans with mental illness, particularly veterans with substance use disorders or serious mental illness who appear to be at higher risk for worse outcomes. This work will likely require ongoing coordination of care with specialty mental health services, as has been examined in other public health care systems (34). Such enhancements may improve care quality of colorectal cancer screening and chronic disease management care for diabetes and ischemic heart disease, indicators where we found greater differences among veterans with a substance use disorder or a serious mental illness.

Although this study examined comprehensive prevention and chronic disease care indicators among individuals diagnosed as having a range of mental illnesses, limitations existed. After consultation with VHA subject matter experts, a cutoff of ≥5% was used to define meaningful differences, which is different from statistical methods used in prior studies with similar aims (17–20). Specifically, multivariate models have been used in previous VHA studies examining similar research questions. Such models were not included in the present investigation given our goal of examining the overall quality of care provided within PACT irrespective of clinical site or patient characteristics and our efforts to identify minimum clinically important differences when comparing adherence to quality indicators. Use of VHA administrative data involved additional limitations. For instance, use of ICD-9 codes to identify patients, although a common approach, may under detect mental illness, and such an approach does not allow for examination of symptom severity. Finally, this is a study of veterans enrolled in VHA primary care, limiting generalizability of results to veterans receiving care outside of the VHA and to civilian populations.

Conclusions

Contrary to prior studies, we found few disparities in preventive and chronic disease management care when comparing veterans with and without mental illness enrolled in VHA primary care services, suggesting that the VHA’s new patient-centered medical home approach to primary care delivery, including integrated mental and behavioral health services, is effective for those with mental illnesses. To further improve care and associated health outcomes, additional efforts to enhance services for veterans with substance use disorders or serious mental illness should be considered.