Introduction

Depression is the leading cause of disability and is one of the most common mental illnesses in the United States [1]. While many people with behavioral health concerns have encounters with their primary care providers [2], healthcare professionals in primary care not using validated screening instruments will only identify approximately half of their depressive patients [3, 4] Assessing depression using standardized questionnaire instruments in primary care has been proven effective in initiating mental health care [5]. There is a growing body of evidence indicating that mental health screening decreases clinical morbidity in individuals with depression due to early treatment in response to positive screening results [6]. Early diagnosis and management additionally improve patients’ quality of life, decrease healthcare costs, and reduce exacerbations of co-morbid medical conditions [7–9]. The United States Preventative Service Task Force (USPSTF) recommended screening for depression in primary care settings for adults in 2016 [10]. A recently released update now also recommends screening for clinically relevant anxiety symptoms [11]. A large study showed increased depression screening rates and increased rates of depression diagnosis and treatment after the introduction of a screening regimen [12]. Particular racial and ethnic minorities and uninsured patients tend to benefit from rigorous mental health screening and subsequent treatment [13]. Although the USPSTF recommendation has been in place for 7 years, many primary care providers still perform mental health screening on a “case-by-case” basis rather than consistently [14], and screening rates have been shown to be 59% [12], 67.2% [15], and 88.8% [16] in various clinical settings.

While there is a growing body of evidence highlighting the benefits of regular mental health screening in primary care, studies of sociodemographic factors associated with lower odds of screening have mostly been conducted before [17, 18] or a few years after the release of the first USPSTF recommendations in 2016 [12, 15, 16]. Newer large-scale data on mental health screening prevalence, which reflects the mental health burden associated with the COVID-19 pandemic [19] is lacking. Identifying these factors is important to inform clinical interventions to improve screening and potential diagnosis. This study addresses this in a large cohort of patients.

Methods

This is a retrospective cohort study of 18 primary care clinics in the Corewell Health healthcare system in West Michigan. This study received approval from the institutional review board of Corewell Health and was deemed non-human subject research (Decision #: 2022–342). The report of the study results adheres to the STROBE guidelines for observational studies [20].

Results

After excluding patients with encounters at multiple primary care clinics and clinical resource encounters, 126,306 patients with 291,789 encounters were included for analysis (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flow chart of patient inclusion.

https://doi.org/10.1371/journal.pone.0301125.g001

Those patients were between 16 and 104 years old (mean: 50.8, SD 18.9), more often of female sex (57.6%) and predominantly White or Caucasian (85.7%).

The primary care clinics were mainly located in metropolitan areas (83.3%) and provided care to an average of 7,017 (min 816, max 23,288, SD 5,613) unique patients during the 14-month study period. Six clinics were situated in counties/municipalities without a mental health provider shortage, while 4 and 8 clinics were located in areas with a moderate or significant mental health provider shortage, respectively. More detailed data are included in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Sociodemographics of n = 126,306 patients.

https://doi.org/10.1371/journal.pone.0301125.t001

On an encounter level, the odds of having any documented PHQ-4/PHQ-9/GAD-7 screening was 66.2% (SD 0.47). On a patient level, 79.8% (95% CI, 79.6–80.0) had a documentation of any PHQ-4/PHQ-9/GAD-7 screening during the 14-month period. 14.3% (n = 14,444) of patients with a documented PHQ-4 screening during the study period were also screened with a GAD-7. Similarly, 10.6% (n = 10,718) of PHQ-4 screened patients received an additional screening with the PHQ-9.

The patient level 14-month screening prevalence (any PHQ-4/PHQ-9/GAD-7) ranged between 51.3% and 98.6% among the included clinics.

The majority of patients had only one (29.3%) or two (28.1%) encounters. Screening prevalence for patients having a single encounter during the 14-month study period was 67.8% (95% CI, 67.5–68.2), for those with two encounters 83.4 (95% CI, 83.0–83.8).

The 14-months screening prevalence rate was lowest among 34-year-olds (75.7%, n = 1,940) and highest among 87-year-olds (88.4%, n = 466). Screening rates were above average among 16 to 18-year-olds and–with few exceptions–among people aged 60 and older (S1 Fig).

In unadjusted analyses, odds of receiving a screening within 14 months were higher for patients who were males (80.1%, 95% CI 79.8–80.5), had an annual physical appointment (81.2%, 95% CI 81.0–81.5 vs. those who did not 76.6%, 95% CI 76.2–77.1) and Black or African American patients had higher screening rates (86.2%, 95% CI 85.3–87.1) vs. White or Caucasian patients (79.1%; 95% CI 78.9–79.4). Patients with up to two appointments had lower (73.9%; 95% CI 73.6–74.2) than average chances to receive mental health screening. Patients with no chronic conditions (diabetes mellitus, chronic ischemic heart disease, chronic inflammatory condition, cancer diagnosis or mental health problems) had 14-month screening rates below average. More information is provided in Fig 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Unadjusted 14-months mental health screening prevalence with corresponding 95% CI.

^a Missing sex n = 4, ^b Includes Native Hawaiian or other Pacific Islander, ^c Missing race / ethnicity n = 1,605.

https://doi.org/10.1371/journal.pone.0301125.g002

Two different multivariable logistic regression models including a total of n = 124,698 patients were performed. All aORs are depicted in Table 2. Model 2 accounts for frequent encounters, while model 1 does not. Both models revealed sex differences on the outcome “Documented mental health screening (any PHQ-4/GAD-7/PHQ-9) within 14 months” in favor of male patients. Also, Black or African Americans, Asians, and Latinas/Latinos/Latinx showed significantly higher odds to be screened compared to White or Caucasians in both models, respectively. Compared to patients with private insurance, those insured through Medicaid were more likely to have a screening documented. Furthermore, patients having at least one annual physical / wellness visit / well child visit had higher screening odds compared to those who did not have one of these preventive encounters.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Results of binary logistic regression models (n = 124,698 patients).

https://doi.org/10.1371/journal.pone.0301125.t002

Patients’ aged 20 to 49 were less likely to have a documented mental health screening in model 1 compared to the reference group of 60-69-year-olds, but when further adjusted to frequency of encounters the effect size decreased and remained only statistically significant for 20-29-year-olds.

The association between clinic size/location and mental health screening prevalence shown in descriptive analyses was revealed in both regression models. Smaller clinics were more likely to perform and document mental health screening compared to larger clinics, and clinics located in areas with a mental health provider shortage performed those screenings significantly more often on their patients. While patients with a mental health diagnosis showed increased screening rates in both regression models compared to those without such a diagnosis, screening prevalence did not differ among patients with diagnosed diabetes or chronic ischemic heart disease, when adjusting for frequent encounters. For cancer patients and patients with chronic inflammatory or rheumatic conditions, screening odds were even lower compared to those without such conditions when adjusted for frequency of encounters.

Discussion

While early detection of mental health disorders has been shown to enhance patients’ quality of life, reduce healthcare costs, and subsequently decrease complications from co-morbidities, our study suggests that providers struggle to screen patients with only occasional encounters, younger patients in their 20s, female sex, White and Caucasian race, having private insurance, and not having preventive annual physical visits. This is concerning as the highest prevalence rates for depression are found among people in their 20s, with women particularly affected at this age [34].

While 14-month screening prevalence in our study was 79.8%, other researchers found considerably lower depression screening rates of 48.6% using a representative sample of US adults aged 35 or older [18]. That study, however, used patient self-reported data and found other demographic factors correlated with mental health screening than in our study. Another study reported similar mental health screening rates, with higher odds in patients aged 65 to 84 and 85+, compared to those aged 18 to 44, however, insurance type was not considered in that study [35].

Our study revealed significant screening rate variations among clinics, despite the health system having a longstanding standard procedure for mental health screening. This heterogeneity in one geographically bounded health system is surprising and concerning. Implementing mental health screening in the work routine was considerably more successful in some clinics than in others. This raises not only questions on structural and organizational barriers that may contribute to under-screening in those clinics but also indicates that a single top-down implementation effort may not be effective among some clinics which may require some more tailored and continuous approaches, e.g. constant monitoring and feedback, incentivizing of screening, highlighting the importance of screening in team meetings, education of office staff, reminder through EMR software, etc. To our surprise, smaller clinics and clinics in areas with mental health provider shortages had substantially higher mental health screening odds. While this phenomenon has to our knowledge not been described before, it is likely that providers in communities lacking mental health providers and with fewer options to refer patients with mental health needs may be more aware of their patients’ mental health problems and perhaps even more keen in treating them. Conversely, providers in areas with sufficient mental health services may be more prone to expect their patients with mental health problems to seek treatment by specialists and thus do not prioritize mental health screening. While scarcity of mental health services is expanding in the U.S. [36], further qualitative research can provide insights on how primary care providers in both provider shortage areas and sufficiently covered areas perceive their professional roles when treating mental health problems. Prior research indicated barriers to primary care providers to screen for mental health, such as lack of training on mental health care [37], anticipated negative reactions to screening from patients, perception of “knowing a patient personally” [38], time constraints [14] and low comfort levels of primary care providers’ comfort in treating psychiatric conditions [39].

Higher screening rates in the elderly may be attributed to Medicare Annual Wellness Visits (AWV) introduced in 2011, offering preventive services that include mental health screening without patient co-payments [32]. The established practice of well child visits, which are regularly offered until the age of 21, and also include screening for mental health, could have also a positive effect on screening prevalence [40]. Unlike AWVs, preventive encounters (e.g. routine annual physicals) in private health insurance plans may either not mandate mental health screenings or propose it but not enforce actual implementation, which may explain lower screening odds.

Furthermore, providers can be prone to bias about who they think may “need” screening, and providers may not be aware of racial and cultural differences in the symptom presentation [41]. However, our study found significantly higher screening rates among Black and African American, Asian and Latinx patients compared to Whites and Caucasian. While this may reflect increased efforts of diversity equity and inclusion initiatives to address mental health disparities in minority populations, other researchers highlighted that implicit bias is a possible explanation that Black people have higher rates of being diagnosed with certain mental diagnoses [42]. Other studies also highlighted lower screening rates among some limited English language proficiency patient groups [16, 43].

Patients with chronic conditions that are associated with higher mental health burdens [28–30] did have higher mental health screening prevalence in unadjusted analyses, but average or even lower odds of receiving mental health screening when adjusted for frequent visits suggesting that encounter frequency is a modifying factor. While our study could show this tendency for various conditions, patients with a known mental health diagnosis were more likely to receive a screening in both regression models.

Quality improvement projects introducing standard of care procedures, treatment algorithms, and/or modified office protocols have shown to increase mental health screening [16, 35, 44]. While these efforts remain on a local level, our study suggested that annual wellness visits correlate with increased screening rates. Encouraging and incentivizing wellness visits, along with incorporating mental health screening as a standard of care procedure, could have a nationwide impact. Timely diagnosis and treatment of depression and anxiety symptoms are beneficial in terms of prognostic outcomes, can prevent worsening of comorbidities, and have been proven cost-effective [7, 8, 45].

Conclusions

This study highlights clinic size, mental health service availability in the community, and number of encounters of a patient are main determinants for receiving the recommended mental health screening. Primary care providers offer accessible mental health services by assessing and diagnosing problems, provide treatment or referrals if necessary, and support patients in managing their condition. As a prerequisite, providers need to be well trained, aware of their important role and comfortable in treating patients. Mental health screening was performed more often by providers of smaller clinics with scarce mental health treatment resources in their communities, suggesting that these providers were more likely to fulfil this role. On the contrary, healthy and younger patients with less healthcare encounters seeking care in larger urban clinics in areas with generally good availability of mental health services were less likely to receive screening for anxiety and depression. While other factors like sex, race or ethnicity also play a role, this finding is particularly concerning given the high rates of depression and anxiety in these age groups.

Furthermore, our study showed a considerable variety of mental health screening prevalence between clinics. This highlights the necessity for a continuous and tailored implementation effort to foster equitable mental health care.

Supporting information