Introduction

Proper provision of mental health services can mean the difference between life and death. Suicide is on the rise and is now the tenth leading cause of death for Americans and the second leading cause of death in the 15-to-24-year age demographic. Suicide is also a significant concern in the Military Health System (MHS). Roughly half of service members are under 25 years of age, and over 70% are male, both of which are demographics with increased risk. Primary care physicians (PCPs) have long been the gatekeepers of the mental healthcare system, and they have a unique opportunity to treat disorders associated with suicide risk. In retrospective studies of patients who committed suicide, 38% visited a provider within the week before they committed suicide, and 64% visited a healthcare provider within 4 weeks before the attempt, with most of those encounters occurring in the primary care setting (Ahmedani et al. 2015). Despite the need for PCPs to identify and treat mental health disorders, detection of mental health conditions in primary care offices is frequently missed, as it is estimated that less than half of all depressive disorders are recognized by PCPs (Egede 2007).

Primary care physicians have become increasingly important to national mental healthcare, as over one-third of Americans live in regions experiencing mental healthcare shortages (Hubbard 2021). In the National Ambulatory Medical Care Survey, the proportion of primary care visits among US adults involving a mental health primary diagnosis increased from 3.4% of visits in 2006–2007 to 6.3% of visits in 2016–2018. The percentage of visits that addressed a mental health concern, not limited to primary diagnoses, also increased over this period from 10.7% to 15.9% (Rotenstein et al. 2023). In civilian healthcare settings, 50–60% of mental healthcare provision occurs in primary care offices, and 79% of antidepressant medication prescriptions are written by non-mental health professionals (Barkil-Oteo 2013). Additionally, PCPs report greater difficulty obtaining mental health services for their patients than for any other specialty, likely due to either a shortage of psychiatric providers or issues with insurance coverage (Cunningham 2009). As most Americans receive their mental healthcare treatment in the primary care setting, research must focus on ways to improve recognition and treatment of mental health conditions in primary care settings.

Mental health conditions comprise a large portion of overall disease burden in the United States. Between 1990 and 2019, Americans lost approximately 7.3 million disability-adjusted life years (DALYs) due to mental disorders (GBD 2019 Mental Disorders Collaborators 2022). Anxiety disorders and depressive disorders are the most common mental health problems in the United States, and patients with these conditions tend to have higher healthcare utilization costs, decreased workforce productivity, and increased disability (Combs and Markman 2014). In addition to the clinical and social cost, the military also has readiness concerns; mental health has been the number one reason for evacuation from overseas since 2013, and in 2020 accounted for nearly one-third of all evacuations (Armed Forces Health Surveillance Branch 2021). Despite new pharmacological, procedural, and therapeutic advances, the disease burden of mental health conditions has remained relatively stable over the past 30 years. In the civilian population, approximately 20% of Americans are affected by a mental health condition during a given year (National Institute of Mental Health 2023).

Among active-duty service members, the prevalence of mental health disorders was approximately 15% in 2018 (U.S. Army Public Health Center 2020). Despite the significant impact of these disorders, to our knowledge, a definitive review of mental health disease burden in MHS care settings has not been completed. In this paper, we intend to use MHS healthcare databases to estimate the number of primary care visits related to mental health conditions, as well as the quantity of psychotropic medications prescribed in primary care settings. These two surrogates should give us a reasonable estimate of the overall mental health disease burden in the DoD and determine what proportion of mental healthcare is being delivered by military-affiliated PCPs. These data will be used as a baseline for future research looking into graduate medical education (GME) training in MH as well as PCP comfort in treating MH. Our goal is to establish data which can be used to improve healthcare delivery and improve active-duty service members and their family’s health.

Methods

This protocol was reviewed by the Walter Reed Institutional Review Board (IRB; WRNMMC-EDO-2022–0890) and was exempted. The MHS Management Analysis and Reporting Tool (M2) is a relational database using web-based technology and contains the most recent five full fiscal years of the Defense Health Agency (DHA) corporate healthcare data. Retrospective analyses of the MHS M2 database from 2017 to 2021 queried frequencies of primary care visits in direct care (DC), namely care received at a military healthcare facility, and in purchased care (PC), care received in the community provided by local civilian healthcare clinics/hospitals. Providers in DC have specialized understanding of the treatment of service members in general, and particularly regarding issues surrounding unique military regulations involved with certain diagnosis and medication classes. Mental health diagnoses were defined by relevant International Statistical Classification of Diseases (ICD) codes in the first through third diagnosis position [a supplemental table with included codes is available upon request from the corresponding author].

Separate queries identified total counts of behavioral health provider visits in DC or in PC over the study period and by year. These data allowed qualitative comparison of trends in frequencies of mental health-related primary care visits with trends in volume of care provided in behavioral health settings. The analysis population was defined as all MHS beneficiaries ages 18–64 years. Primary care visits and behavioral healthcare visits were defined by the DoD Medical Expense and Performance Reporting System (MEPRS) or provider type and excluded DC visits with administrative codes. Denominators for stratified analyses by patient characteristic were defined as the total number of primary care visits or prescription fills with that characteristic. Frequencies of mental health services in primary care were described across strata as counts and prevalence with their 95% confidence intervals (CI). Prevalence ratios were compared with Poisson regression models by year and by characteristic, with potential linear trends evaluated over the study period. Models defined the outcome as the count of mental health-related visits, with continuous year and beneficiary category or DC versus PC as independent variables and strata denominators as the offset.

The Pharmacy Data Transaction Service (PDTS) is a central repository for prescription data from all DoD pharmacy services. PDTS uses a centralized data system to store and analyze information about prescriptions filled across the DoD’s network of pharmacies. A limitation of the PDTS system is that ordering provider type is not classified for PC prescribers; therefore, the analysis was limited to prescriptions ordered in DC settings. Frequencies of filled prescriptions for mental health-related drug classes (i.e., psychotropics) in DC were queried over this study period from the PDTS for primary care ordering providers. Medication designations included anticonvulsants, psychotherapeutic agents, antidepressants, amphetamines, anxiolytics, barbiturates, benzodiazepines, sedatives, hypnotics, antimanic agents, and alcohol deterrents.

Results

The MHS beneficiary population among those aged 18–64 years in 2021 was 53.5% male, 31.1% active duty (AD), 15.5% dependents of AD, 20.9% retirees, and 32.5% other. These beneficiaries had 148 million total outpatient visits from 2017 to 2021 in DC and 217 million in PC. A total of 64.0 million DC visits were in primary care and 13.0 million were in behavioral health; 46.4 million PC visits were in primary care and 14.5 million in behavioral health (Fig. 1). Primary care visits in DC were more likely to be from younger and AD beneficiaries versus primary care visits in PC. The MHS has a codified priority to treat AD in DC, which likely explains why most of the AD remain in DC while most of the dependent care is PC. The prevalence of mental health disorders in AD service members has been reported to be around 15% based on relevant ICD codes from DC and PC visits including to behavioral health providers (U.S. Army Public Health Center 2020).

Fig. 1
figure 1

MHS visits by year

Full size image

The total count of visits to behavioral health providers in DC decreased from 2,876,529 in 2017 to 2,337,271 in 2021, a 19% decrease over the 5-year period. In PC, this total count increased from 2,513,901 in 2017 to 3,538,712 in 2021, a 41% increase over the same period. For AD, the number of DC behavioral health visits in 2017 was 2,514,071, and in 2021 that number fell to 1,942,111 (a 22.8% decrease) while PC increased from 268,640 to 587,618 (a 119% increase). In 2017, 90.3% of AD BH care was provided by DC, while by 2021 the amount fell to 76.8% (Fig. 1).

The overall percentage of primary care visits in either DC or PC with a mental health diagnosis was 7.3% over the period 2017–2021 (Table 1). Annual percentages increased from 7.0% in 2017 to 7.9% in 2021. The overall percentage was 4.3% for AD, 6.1% for retirees, 11.9% for dependents, and 9.5% for all others, primarily including dependents of retirees; and 5.2% in men and 9.3% in women. For AD the percentages were similar across branches of service.

Table 1 Primary care visits
Full size table

In DC, 5.7% of primary care visits had a mental health diagnosis in both 2017 and in 2021 (Fig. 2). The overall percentage over the study period was 3.9% for AD versus 11.6% for dependents (Table 1). In PC, 8.9% of primary care visits had a mental health diagnosis in 2017 versus 10.5% in 2021. The overall percentage was 8.4% for AD versus 12.4% for dependents.

Fig. 2
figure 2

Percent of primary care visits resulting in a mental health diagnosis

Full size image

In Poisson regression models of prevalence ratios for mental health visits by year and military status or location of care, dependents of AD had 2.97 times higher percentage of DC and 1.48 times higher percentage of PC primary care visits with a mental health diagnosis, relative to AD, over the study period. Additionally, relative to DC primary care visits, PC primary care visits for AD had 2.14 times higher percentage and for dependents 1.06 times higher percentage of visits with a mental health diagnosis. All reported prevalence ratios were significantly different from 1 (p < 0.001 for each). Qualitative linear increases in the percentage of primary care visits with a mental health diagnosis were observed across study years in PC but not in direct care (Table 2).

Table 2 Primary care prescriptions
Full size table

Of PDTS prescriptions ordered in DC by a primary care provider, 9.6% of prescription fills had a mental health-related drug class in 2017 and 10.0% in 2021 (Fig. 3), with overall percentages of 7.7% for AD versus 15.5% for dependents. For AD this percentage increased from 7.6% in 2017 to 9.3% in 2021, and for dependents increased from 15.2% in 2017 to 16.7% in 2021. No qualitative linear trends in the percentages were apparent across study years.

Fig. 3
figure 3

Percent direct care psychotropic prescriptions

Full size image

Discussion

This study found several areas of agreement and divergence from previous data which merit further discussion. In a study looking at over 8600 unique psychotropic prescriptions in civilian populations over a decade, no psychiatric diagnosis was recorded for over 60% of visits at which a new psychotropic prescription was initiated (Rhee and Rosenheck 2018). Another study of over 5 million patients found that 58% of individuals who were prescribed a psychotropic medication in 2009 had no psychiatric diagnosis at any point during the year the prescription was filled (Wiechers et al. 2013). This current review found that there was roughly twice the percentage of AD visits in which a psychotropic medication was prescribed as there were visits resulting in a psychiatric diagnosis (7.7% vs. 3.9%), while the dependent group remained more closely aligned (16.5% vs. 15.5%). While there are legitimate non-psychiatric uses of antidepressants (migraine, fibromyalgia, menstrual symptoms, etc.), another study of over 4000 physicians found that less than 8% of psychotropics were prescribed for a non-psychiatric condition, a number which is inadequate to explain the difference (Mark 2010).

From these study data, it seems that primary care in the MHS is treating somewhat less behavioral health than their civilian counterparts, which in 2018 was estimated to be roughly 15% of primary care visits (Armed Forces Health Surveillance Branch 2021). This may be due to the more seamless ability to engage behavioral health within the MHS or to a lack of comfort treating this population leading to earlier referral to behavioral health. Interestingly, the likelihood of receiving a mental health diagnosis in primary care was similar between DC and PC (11.6% vs. 12.4%) in the dependent population, while it was more than twice as likely that an AD PC visit would result in a mental health diagnosis as a DC visit (8.4% vs. 3.9%). While these discrepancies are not unique to the MHS system, there could be additional reasons DC providers treating the AD population might be hesitant to report psychiatric conditions such as fear of possible effects on the service member’s career or deployability. This study was not designed to definitively speak to this discrepancy, and further research is warranted. There are some other likely possibilities discussed here. PC providers who see AD patients are not always familiar with the intricacies of DoD instructions related to medical readiness and may therefore be more likely to make the diagnosis earlier than the DC provider. This lack of diagnosis by DC providers in AD populations could be comparable to PC providers concealing psychiatric diagnoses from patients’ notes for prevention of stigma or adverse legal or occupational consequences, or insurance considerations. It is also worth noting that the AD population tends to be younger and healthier than the civilian population, and as such may have forms of anxiety or mood disorders that do not fully meet criteria for a Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) diagnosis. Americans in general are also more receptive to psychiatric medication in recent years, so it is possible that patients with milder conditions may be more open to psychiatric medication compared to a time when psychotropics were more stigmatized.

Another noticeable trend was the movement of both AD and dependent behavioral healthcare from DC to PC. Most of the AD primary care and BH care was provided by DC, which is consistent with DoD triage priority of AD care and a general greater willingness to defer non-AD to PC. Between 2017 and 2021, the number of DC BH visits in AD declined by ~571,000 (22.8% decrease). The concomitant increase of ~319,000 PC BH visits (119% increase) in AD represented only 56% of this substantial decline in DC BH visits. While this study does not have the granularity to determine a definitive reason for the decline in visits, national trends make it unlikely to be related to decreased demand. In dependents, like the qualitative trends in AD, DC BH visits declined by ~85,000 (46.7%) over this period while PC BH visits increased by ~408,000 (59.3%). Primary care visits, in contrast to the decline in BH, remained level across years for AD in DC, but decreased ~30% for dependents in DC, and increased ~30% for both AD and dependents in PC.

Conclusions

This study is the first to our knowledge to use large, comprehensive MHS healthcare databases to characterize more than 110 million unique encounters to determine behavioral health needs in primary care among military service members and their family members.

When compared to national statistics and noting the abovementioned discrepancies, it is likely that our observed counts of mental health-related visits defined by diagnosis codes represent undercounts of all MHS primary care visits that provided care for a mental health concern. However, based on these data, we can conservatively posit that the typical primary care provider working in the MHS can expect 7–10% or more of their encounters to have a behavioral health component. To assess the potential need for primary care training to adapt to this demand, this group is currently evaluating whether clinicians are adequately trained to handle this need and how comfortable MHS providers are in addressing these conditions, and plan to present these data in a later publication.

The study is limited in that no attempt was made to look at individual records to determine the diagnosis or reason for lack thereof; a future study might evaluate a representative sample of encounters to better elucidate these reasons. Further, the authors have yet to fully analyze the data to break down the types of diagnosis being treated, or the preponderance of medication classes being prescribed in primary care beyond relevant ICD and drug class; however, this work is in data analysis for a future manuscript. Future studies should further characterize the discrepancies in MH diagnoses for AD between DC and PC. This study is the first step in a multi-part research project with a long-term goal of evaluating potential barriers to care and improving behavioral healthcare provision for AD service members and their dependents.