Background

Some veterans using healthcare services at the Department of Veterans Affairs (VA), face multiple potential barriers to VA healthcare services, particularly mental healthcare [1, 2]. Previous research has examined the barriers faced by veterans obtaining needed VA mental healthcare services [3,4,5,6,7,8,9]. Limited attention has been paid to the influence of veterans’ experiences (e.g., prior military service, day-to-day struggles) and meaning systems on their perception of barriers to VA mental healthcare services.

Patients, physicians, and researchers who have different life experiences and backgrounds may use different paradigms and language—that is, different cultural frameworks when thinking about healthcare or engaging with healthcare systems [10,11,12]. For example, veterans’ military history greatly influence their understanding of mental health and the language used to describe those experiences [13], while physicians’ and researchers’ biomedical and scientific research training shapes their understanding of health and use of language to describe patient experiences [14]. The VA has made important strides toward veteran-centric care that places the voice of veterans and their unique needs at the center of healthcare. Recently, the VA has implemented and funded a number of initiatives (e.g., Center for Compassionate Innovation, MyVA, Whole Health Initiative), as well as research, trainings, and quality improvement projects to facilitate this culture shift [15,16,17].

In this era of greater emphasis on patient-centered care approaches, where researchers are encouraged to meaningfully collaborate with populations they study [10], one critical care need is to understand the barriers to care grounded in veterans’ frameworks for thinking about health. Veterans with mental health problems tend to perceive more barriers to mental health treatment than veterans without similar problems [18,19,20,21]. By understanding veteran-centric barriers to mental healthcare, we can better understand how to increase initiation and engagement with VA mental healthcare services, as well as adherence to mental health treatments.

In this paper, we report on the use of cultural domain analysis (CDA), a method common in ethnographic research [22], to elicit veteran-centered concepts of barriers to VA mental healthcare services use. We selected this approach over quantitative methods such as survey questions because it allows veterans to define barriers to care that they themselves experience, and to talk about how their experiences influence their mental health service use. Using a series of elicitation and interpretation interviews, we elicited veteran-centered barriers to care and then asked veterans to explain the meaning of these barriers in their own VA healthcare experiences. CDA methods elicit an “emic” perspective (the perspective of the participant), rather than an “etic” perspective (the perspective of the researcher) [23, 24]. This analysis is the initial phase of a sequential mixed-methods study designed to develop and validate a perceived access measurement tool. The first phase of the larger study involved qualitative semi-structured interviews with veterans, of whom a subsample participated in the CDA and domain interpretation interviews. This paper reports on these findings.

Methods

We used standard CDA procedures to elicit veteran-centered barriers to mental healthcare and to create a cognitive map [25, 26]. We followed up with qualitative, semi-structured interviews that were conducted during one-hour interview sessions to understand the meaning of these barriers in veterans’ healthcare experiences. For all interviews, participants completed a brief quantitative sociodemographic survey and reported on their current mental health symptoms severity for depression, post-traumatic stress disorder (PTSD), and alcohol use. Depression severity was measured using the Patient Health Questionnaire nine-item depression module (PHQ-9) from the PRIME-MD [27,28,29]; PTSD symptom severity was measured using the PTSD Checklist Civilian version (PCL-C) [30]. At-risk drinking was assessed using the AUDIT-C [31]. All participants provided written informed consent and were remunerated for their study participation. Prior to the start of the research, the VA’s Central Institutional Review Board approved all study procedures.

Setting

This study was conducted at ten sites in three geographically distinct VA administrations areas called Veterans Integrated Service Networks (VISNs): VISN 1 included one urban and two rural CBOCs in Maine serving primarily white veterans; VISN 16 included one urban health clinic and two rural CBOCs in Arkansas serving a largely white and African American veteran population; and VISN 21 included one urban and three rural CBOCs in Northern California serving a racially and ethnically diverse group of veterans. Table 1 shows the rural, urban, and ethnic/racial breakdown across sites.

Table 1 Free list items within their dimensions of the cultural domain
Full size table

Recruitment

We recruited veterans who used VA healthcare and had behavioral health concerns, but who did not necessarily use specialty mental health services. Eligible participants included veterans between 18 to 70 years old who screened positive for at least one behavioral health problem, including depression, PTSD, or at-risk alcohol use during a routine VA primary care appointment within the past year. We used opt-out letters, a proven strategy for gaining veterans’ participation in health services research [32,33,34], to recruit a diverse sample of veterans across the three states.

As described in detail elsewhere [35, 36], we sent opt-out letters to anywhere from 30 to 60 eligible veterans within each of the three VISNs repeating this procedure in three successive waves. Opt-out letters were sent to 585 veterans across all three sites. Eligible veterans received a letter describing the study and the opt-out process that involved returning a self-addressed, stamped, “do not call” form or calling a 1–800 number to opt-out. A total of 89 veterans opted out of the study. Within two weeks of receiving the letter, study staff called the remaining 496 veterans who did not opt-out. Veterans who had access to a telephone and had experienced mental health problems related to PTSD, depression, or alcohol use were invited to participate in the study. An additional eight veterans had already been recruited into the study during a pilot phase to assess the study design. We called to confirm eligibility and discuss study participation; 258 veterans were reached and 72 participated in the study. The final study sample included 80 veterans.

Based on preliminary analyses and review of our recruitment log, we then purposively identified certain subpopulations of veterans (e.g., rural veterans, Iraq and Afghanistan veterans, women veterans) to recruit, ultimately diversifying our sample. In wave 1, we recruited veterans of differing service eras obtaining representation of older and younger veterans (e.g., Vietnam veterans, Iraq and Afghanistan veterans); in waves 2 and 3 we recruited veterans based on race and ethnicity. For all waves, we recruited veterans from both rural and urban contexts and over-recruited women veterans, an underrepresented veteran subpopulation.

Of the 80 veterans in the sample, 66 veterans participated in the cultural domain analysis and interpretation interview. As described below, this method requires a relatively small sample size. Therefore, we recruited participants into the CDA phase until agreement was reached (n = 52) and into the domain interpretation task until saturation was reached (n = 14).

Cultural domain analysis and interpretation interview

Cultural Domain Analysis (CDA) examines how members of a group who share a culture characterize aspects of that culture through a cognitive domain, referred to as a cultural domain in CDA, defined by a set of words, phrases, or concepts that collectively symbolize a single idea [37]. The CDA method elicits participants’ words and brief phrases that constitute a domain, analyzes its structure, and assigns those structures into one or more dimension. CDA is a multi-step, iterative mixed-methods approach that intersperses data collection with analysis. Analysis of each step systematically informs the next step to ensure a comprehensive interpretation. Sample size determination for the CDA steps is based on inter-correlations among participants about domain items and requires very small sample size; and high domain agreement can be reached with as few as 15 participants per step [38, 39]. We added a final phase of qualitative interviews to validate domain dimensions and elicit veterans’ understandings of the meanings of identified barriers in veterans’ healthcare experiences [40]. We describe each step of the CDA process below.

Free list exercise. We started with a free list exercise, an unstructured, open-ended task used to elicit the words and brief phrases that inductively characterize a domain. During data collection, each participant was asked to “list all the things that you think can make it more difficult for veterans to get help for stress-related and emotional health problems (see Additional file 1 for a copy of the guide). Participants listed as many items as they could in response to the verbal prompt. To elicit additional items, interviewers used common free list probing techniques, including non-specific prompts, such as “What other kinds of things make it more difficult to get help?”; reading back free lists to check for accuracy, followed by non-specific prompts; and semantic cues, such as “What about [transportation], can you think of anything else that is similar to this item?” [41]. A total of 23 veterans across the three VISNs participated in the free list task.

Participants’ responses included a total of 238 unique single-word items and brief phrases. The qualitative team reviewed the items, eliminated duplicates, and combined similar items to reduce the list; calculated the frequency of items; and selected items with a frequency ≥ 5 and items with lower frequency but high importance based on qualitative analyses of interview data [42]. Table 1 shows the master list of 46 unique items generated by this data reduction process and used in the subsequent pile sort activity.

Pile sort activity

After compiling the free lists, we asked a new set of participants to conduct a pile sort activity in which participants organize or sort a finite number of discrete words and phrases into categories according to whatever criteria they see fit. The 46 items generated in the free list exercise were placed on index cards and assigned a unique number. Interviewers asked participants to “place all the items you think are similar into piles,” using as many or as few piles as necessary. Finally, participants were asked to review and label each pile with a descriptive title. A total of 29 veterans across the three states participated in the pile sort activity.

Pile sort data were entered into a database keyed on index card label and unique number and were imported into Anthropac, an analytic program designed to collect and analyze cultural domain data by identifying underlying relations between items and meaningful group clustering [43]. Pile sort data were analyzed using the Anthropac multidimensional scaling (MDS) tool to identify each statement as a separate point on a two-dimensional map, which provides a visual representation of participants’ sorting data as illustrated in Fig. 1 [44]. MDS graphs prioritize meaning through spatial proximity (i.e., items appearing closer together are more conceptually similar; items appearing farther apart are more conceptually distinct). MDS calculates a stress value, a diagnostic statistic measuring the overall fit of the map (i.e., goodness-of-fit indicator). Lower stress values more accurately reflect congruence between the raw data and the computed distance matrix. For example, 0.0represents perfect congruence between the raw data and the distance matrix whereas ˃0.2 represents poorer congruence and goodness of fit [45]. The average estimated stress value for similar studies using MDS to analyze pile sort data is .280 [46]. We obtained a stress value of .181, representing a good fit of the map to the data. Finally, we conducted a cluster analysis of the MDS to identify meaningful groupings of items across participants as shown in Fig. 2 [47].

Fig. 1
figure 1

Multidimensional scaling of free list items

Full size image
Fig. 2
figure 2

Cultural dimensions of access barriers

Full size image

Domain interpretation task

Finally, after generating MDS and cluster analyses, we asked a third set of participants to examine the results as a means to validate the domain and its dimensions and understand the meaning of the dimensions and constituent items in veterans’ VA healthcare experiences. We asked participants to explain why items clustered together and their relationship to veterans’ healthcare experiences. We then asked veterans to label each pile with a unique descriptive title they created to characterize the pile. This resulted in different descriptive titles from the pile sort exercise. A total of 14 veterans across the three VISNs participated in this phase.

Participants’ responses during the interpretation task were audio-recorded, professionally transcribed, and imported into qualitative data management software [48]. Following the analytic process described by Ryan and Bernard [49], the first author read the transcribed text line-by-line to identity themes, writing memos throughout. A coding schema was developed with code definitions, and thematic codes were applied to narrative text [50]. Axial coding was used to make comparisons across transcripts and to understand the relations among themes within and across transcripts [50]. The qualitative team reviewed the resulting analysis, and the full qualitative team contributed to and decided on final cluster names through an iterative process.

Results

Sixty-six unique veterans between the ages of 20 to 70 (average 44.0) years participated in one of the three CDA phases (See Table 1). Most participants were male (n = 50), white (n = 40), and had prior specialty mental health visits (n = 40); 24 screened positive for PTSD, 28 for depression, and 40 for at-risk drinking. All had experience using the VA healthcare system; consequently, their understanding of barriers to care was grounded in their experiences.

The cultural domain is presented in Fig. 2. Numbers denote the items in the domain dimensions corresponding to the item numbers in Table 2. In MDS, items closer together are more closely related to one another; and in hierarchical cluster analysis, clusters of items are placed together to represent dimensions of the larger cultural domain.

Table 2 Participant characteristics
Full size table

The CDA procedures resulted in five dimensions of barriers to care: worry and concerns about what others think; personal, financial, and physical obstacles; confidence in the VA healthcare system; navigating VA benefits and healthcare services; and, privacy, security, and abuse of services. (These labels represent the most common words and phrases veterans used to label each dimension.) Figure 2 provides a visual representation of domain dimensions and Table 3 provides narrative text to further explicate the meaning of domain items to veterans’ healthcare experiences.

Table 3 Themes in veterans’ interpretations of the five dimensions of barriers to care continuum
Full size table

Dimension 1: Worry and concern about what others think

Veterans’ discussion of this dimension grouped around stigma, vulnerability, and trust as barriers to VA mental health services use. Veterans expressed concern over stigmatizing labels such as “crazy” and “mental health patients” as well as military attitudes (e.g., “suck it up”) fostering feelings of weakness and failure. Such attitudes contributed to veterans’ anxiety and alcohol and drug use and impeded mental healthcare services use. Veterans also indicated a lack of trust in the clinical encounter (e.g., interactions with non-military healthcare providers with limited understanding of veterans’ military experiences) decreased their motivation to remain in care.

Dimension 2: Financial, personal, and physical obstacles

Discussions centered on personal life struggles (e.g., poor health, legal and financial issues, and homelessness) constraining on veterans’ daily lives and ability to meet responsibilities—including healthcare services use. Travel distance to VA facilities and lack of transportation or gas money also represented barriers to VA care within this dimension.

Dimension 3: Confidence in the VA healthcare system

Veterans’ discussions of this dimension indicated system-wide problems with VA processes of care and expressed their lack of confidence (e.g., availability of timely visits, follow-up care). Appointment problems, staffing issues, limited follow-up from VA healthcare providers and staff, and prioritization of some veterans over others impeded access to care. Long wait times between appointments, slow clinic response, and challenges to scheduling appointments (e.g., being passed from clinic to clinic) created barriers to timely and appropriate care. Staff issues, especially lack of available providers and high provider turnover, limited veterans’ access to care and provider continuity over time. Veterans also indicated that limited or no access to specialized care resulted in veterans’ disengagement with care. Furthermore, the appearance that services are prioritized to address the needs of younger veterans limits available services for older veterans and diminishes the importance of older veterans’ health and wellbeing.

Dimension 4: Navigating VA benefits and healthcare services

Veterans’ discussions highlighted the challenges of navigating VA benefits and healthcare services, which are connected to lack of understanding or misunderstanding of VA benefits and mental healthcare services. Veterans reported being unaware of VA benefits prior to leaving military service, having limited knowledge of the processes required to obtain benefits (e.g., necessary paperwork) once eligible, and struggling to understand the use of their benefits as veterans. This knowledge gap impeded initial enrollment into the VA healthcare system. Once veterans entered the system, they still struggled with a lack of awareness of available mental healthcare services. Veterans stressed the need for both the military and the VA to disseminate benefit material more effectively to service members prior to military exit.

Dimension 5: Privacy, security, and abuse of services

Veterans’ discussions of this dimension revealed concerns regarding healthcare privacy and misuse of the VA healthcare system by other veterans. Veterans discussed military policy denying security clearances for service members known to seek mental healthcare and this fear influencing veterans’ decisions to seek specialty care within the VA healthcare system. Some also expressed a fear of their confidential information being shared and sensitive information (e.g., actions in combat, mental health diagnoses) disclosed, which could have profound negative consequences on veterans’ lives (e.g., incarceration, loss of security clearance). Both older (Vietnam era) and younger (Iraq and Afghanistan era) veterans’ expressed distrust of the VA healthcare system as a government institution and fear of digital communications (e.g., tele-psychiatry). Last, misuse of VA services by some veterans for disability payments and pain medication, while perceived by veterans as rare, had a negative effect on the VA care experience of other veterans. Specifically, misuse of the system influences how society and some providers view veterans who use VA healthcare for certain treatments (e.g., pain medication and mental health treatment).

Discussion

This mixed-methods CDA analysis generated veteran-centered perspectives of barriers to VA mental healthcare services use grounded in veterans’ lives and during clinical interactions that add depth, breadth, and complexity to understanding barriers to care. This approach contributes to better understanding of veteran-derived perspectives on barriers to mental healthcare services use. The findings are intended to inform patient-centered care [51] and conceptual models, such as the SOTA Access Model [52], and measurement tools [20].

Our interpretation of veterans’ perceptions of barriers to VA service use demonstrates that historical, socio-cultural, and psychological factors inform veterans’ decisions to engage with VA mental healthcare services. Their perceptions highlight how military socialization, command structure influences, and institutional attitudes (e.g., “suck it up” mentality) reinforce attitudes around help-seeking as weakness, which significantly informs veterans’ post-military service healthcare experiences [7, 53]. The findings suggest veterans and active duty service members experience similar system-level and socio-cultural barriers to use of government mental healthcare services. In their study of barriers to engaging service members in the Department of Defense (DoD) mental healthcare services, Tanielian et al. [54] found concerns over military healthcare system capacity and processes of care (e.g., limited providers, problems scheduling appointments) and the military ethos of “toughing it out” greatly influenced active duty services members engagement with military mental healthcare services.

In addition, our findings show how institutional betrayal, or as veterans explained, distrust of the VA as a government institution, influences their VA mental healthcare services use. Desai et al. [5] found negative perceptions of social and governmental institutions created barriers to seeking VA mental healthcare services among Vietnam-era veterans. Similarly, veterans in our study, who represented Vietnam, Persian Gulf, and Iraq and Afghanistan service eras, distrusted providers’ use of computers to document personal and sensitive information, questioned the security of digital communications (e.g., video-conferencing), and feared hacking of electronic medical records.

This study also provides insight into ways veterans perceive misuse or abuse of the VA healthcare system. The VA is well aware of potential abuse of its healthcare services. For instance, the VA Office of Community Care has a webpage with frequently asked questions about fraud, waste, or abuse of VA healthcare services (https://www.va.gov/COMMUNITYCARE/about_us/POI/poi_faq.asp). Veterans’ abuse of VA healthcare was present in our study—in the case of a participant retelling how veterans scheme VA doctors to obtain pain medication only to misuse their prescription. Whereas fraud was present in the case of a participant explaining that some veterans manipulate the system to obtain a PTSD diagnosis. Though veterans described them as rare, these practices can shape veterans’ and providers’ attitudes and impede development of a trusting therapeutic alliance between providers and veterans utilizing VA healthcare services.

Our findings provide an evidence base for DoD and VA health policy. We found that norms and values embedded within the military perpetuate stigmatizing attitudes around mental health treatment seeking and continue to influence veterans’ healthcare utilization decisions post military service, and that veterans need support transitioning from the DoD to VA healthcare system. This finding reinforces the need for ongoing collaborations between the VA and DoD to critically address military institutional norms and attitudes stigmatizing service member’s mental health-seeking behaviors and facilitate care coordination as active-duty service members transition to veterans.

Policy to reduce stigmatizing attitudes about mental health and treatment seeking in the military and healthcare transition processes that link service members to VA care prior to leaving the service would be ideal. The VA/DoD “One Mission- One Policy- One Plan” (a community of practice) facilitating service member and veteran care coordination is a step towards reducing barriers to care for younger veterans [55]. Similar communities of practice, such as VA and community-based organizations serving veterans, for older generations of veterans who are not using VA benefits and healthcare services are needed [56].

Furthermore, our findings regarding veterans’ confidence in the VA healthcare system points to service members and veterans’ mistrust in government healthcare. Both VA and DoD should work together to address veterans’ mistrust of the VA healthcare system, which reflects both current public relation issues and the historical treatment of veterans of different service eras (e.g., Vietnam veterans vs Iraq and Afghanistan veterans) within the VA healthcare system [57, 58]. Furthermore, as the use of electronic medical records, digital communications (e.g., tele-psychiatry, secure doctor-patient messaging), and smartphone applications evolves through the twenty-first century, veterans’ distrust of the VA healthcare system as a government institution may impede use of technology which is important issue to address for some veterans.

Several limitations should be considered when interpreting the findings. Our findings reflect insight from veterans enrolled in VA healthcare services and may not reflect the experiences of veterans who have not used VA benefits and services. While we had much success recruiting participants from both rural and urban settings across the three sites (Arkansas, California, and Maine), we enrolled a relatively small number of racial/ethnic diverse and women veterans into the study. The general trend of racial/ethnic diversity in our sample, including representation of Asian identity in North California, Black and White identity in Arkansas, and White identity in Maine and Northern California, corresponds with the demographic characteristics of the veteran population in these geographic regions [59,60,61]. Additionally, we purposively sampled across VISNs and combined data across sites; therefore, our sample size is not sufficient to conduct meaningful cross-site comparisons limiting our understanding of potential geographic and racial/ethnic variation.

We oversampled women veterans and enrolled a fair number of women (n = 16) into the study. However, our analysis could have been qualitatively more nuanced and rich if more women had participated in the study and shared their experiences and perspectives. Historically, servicewomen and women veterans have been underrepresented in research examining VA or DoD healthcare use [62, 63]. Women veterans are the fasted growing segment of VA healthcare users [64]; yet, they represent a small population of VA patients and are often underrepresented and potentially excluded from analysis [65]. Use of a context-specific or gender-sensitive sampling approach may have increased women’s participation in our study [65]. Women veterans differ from men in healthcare utilization, how their VA care is organized (e.g., concentrated care in women only clinics), and gender-specific demands (e.g., primary caregivers), thus a one-size fits all approach to recruitment may limit women’s research participation. Future research should engage women’s health program leaders such as medical directors and program managers to identify recruitment strategies that are context-specific and therefore identify how and where women access healthcare and characterize their healthcare utilization patterns, as well as are sensitive to the unique demographics and healthcare experiences of women veterans. Such an approach would likely result in a gender-specific rather than gender-neutral recruitment approach and potentially two or more distinct recruitment strategies.

Finally, some veterans struggled to understand the purpose of the interpretation task and found it challenging to provide explanations for the clustering of items within the five dimensions of the shared cultural domain. After adjusting the instructions to include a brief hypothetical example, veterans more readily understood the task and offered important insight into how barriers within dimensions played out in their or other veterans’ healthcare experiences. Therefore, concrete instructions and an initial example are needed to successfully employ these methods.

Conclusion

Our study findings reinforce the value of eliciting veteran’s perspectives on barriers to VA healthcare service use. By using participatory methods to elicit participants’ perspectives, we obtained veteran-centric barriers to care that are embedded in VA healthcare experiences and offer insight on how to decrease barriers to VA mental healthcare services use. The CDA approach outlined in this paper places patients at the forefront of data generation, analysis, and interpretation, which results in a shared understanding of barriers to care grounded in veterans’ experiences. Empirical, patient-centered data are critical to informing healthcare policy as well as developing valid measurement tools to accurately measure veterans’ experiences of barriers to care; these data can inform instrument domains and provide vocabulary, idiomatic expressions, and scenarios shared among patient populations [66].