Primary Care Physician Recognition and Documentation of Depressive Symptoms Among Chinese and Latinx Patients During Routine Visits: A Cross-Sectional Study

Purpose: Asian and Latinx individuals have a high burden of untreated depression. Under-recognition of depressive symptoms may contribute to existing disparities in depression treatment. The objective of this cross-sectional study was to determine whether physicians recognize and treat depressive symptoms for Chinese and Latinx patients during routine primary care visits. Methods: We analyzed data from 1171 Chinese and Latinx patients who were interviewed within 1 week after a primary care visit in a large academic practice, which had not yet implemented universal depression screening. We included participants with depressive symptoms (defined as a Patient Health Questionaire-2 score ≥3) and no prior history of depression (N=118). We investigated whether patients perceived having a mental health need in the prior year and conducted chart reviews to assess provider recognition of depressive symptoms, defined as documentation of symptoms, antidepressant initiation, or mental health referral within 30 days of the visit. We further examined differences by race/ethnicity and language preference. Results: Among the 118 patients with depressive symptoms and no prior depression diagnosis (mean age 68), 71 (61%) reported a mental health need in the prior 12 months; however, providers recognized depressive symptoms in only 8/118 patients (7%). The number of patients with recognized symptoms was small across race/ethnicity and language preference groups and we found no significant differences. Conclusion: Physicians recognized and documented depressive symptoms for 1 in 10 Chinese and Latinx patients during routine primary care visits. Targeted efforts are needed to address under-recognition of symptoms and improve depression care for these populations.


Introduction
Depression is associated with poor mental and physical health outcomes and earlier mortality. [1][2][3][4][5][6][7] While depression treatment can improve depressive symptoms, as well as management of chronic health conditions, many patients continue to have undiagnosed and untreated depression. [8][9][10][11] Experts have called for the aggressive detection and treatment of mental health symptoms, particularly during times of increased stressors, such as the COVID-19 pandemic. [12][13][14] Yet prior studies suggest that physicians fail to recognize depressive symptoms in a large number of their patients. 11,[15][16][17] Prior studies have found that racial/ethnic minority patients who are clinically depressed are less likely to be diagnosed and treated for their depression compared to non-Hispanic whites. 15,[18][19][20][21][22][23][24][25][26][27][28] Asian and Latinx individuals are more likely to have untreated depression. 18,19,29,30 Yet it is unknown to what extent physicians recognize depressive symptoms among Chinese and Latinx patients without a prior diagnosis of depression and how this may contribute to mental health treatment disparities. Without recognition, physicians cannot begin to address depressive symptoms, representing a missed opportunity for diagnosis and leading to a potential gap in treatment for patients with depressive symptoms. Primary care physicians are uniquely poised to detect, refer, and/or treat patients with depressive symptoms.
A better understanding of gaps in symptom recognition among Chinese and Latinx patients will help to improve depression screening and management in primary care. The primary aim of this study was to investigate the proportion of Chinese and Latinx patients with current depressive symptoms, whose symptoms were recognized during a routine primary care visit in the absence of universal depression screening. A secondary goal was to examine differences in depression symptom recognition by race/ethnicity and preferred language.

Sample
In this secondary data analysis, we analyzed data from the Language Access System Improvement (LASI) study, which has been described previously. [31][32][33][34] LASI was designed to evaluate the effects of simultaneously increasing access to professional interpreters and certifying bilingual physicians' language skills on communication and clinical outcomes in primary care. LASI recruited patients from an academic primary care practice, which serves an ethnically, linguistically, and socioeconomically diverse population of over 25,000 adult patients in San Francisco.
Patients were eligible for LASI if they were ‡ 40 years; self-identified as Latinx or Chinese (the largest populations with limited English proficiency [LEP] in our practice); preferred English, Spanish, Cantonese, or Mandarin; received primary care in the practice; and were able to participate in a telephone interview. In this practice, patients with LEP tend to be older than the general English-speaking population. To avoid confounding by age, LASI purposively sampled older English speakers for a comparison group, resulting in an older mean age of study participants. Bilingual/ bicultural research assistants conducted telephone interviews in the patient's preferred language within 1 week of a primary care visit (the ''index visit''). The majority of LASI participants were seen by primary care physicians, with a smaller number seen by primary care nurse practitioners or medicine residents between January 2016 and July 2017. This study was conducted before the implementation of universal depression screening in the primary care practice.

Current depressive symptoms
We defined current depressive symptoms using the participant's response on the post-visit telephone survey to the Patient Health Questionnaire-2 (PHQ-2). The PHQ-2 is a well-validated and widely used clinical measure to screen for current depressive symptoms, which includes two questions, ''In the two weeks before coming to the clinic, how often were you bothered by any of the following problems? (1) Little interest or pleasure in doing things? (2) Feeling down, depressed, or hopeless?'' Possible answers to each question include ''not at all,'' ''several days,'' ''more than half the days,'' and ''nearly every day,'' or decline to answer, with total possible scores from 0 to 6. Participants were categorized as having current depressive symptoms if they had a PHQ-2 score ‡ 3 (sensitivity of 61-87% and specificity of 78-92% for major depression [35][36][37][38] ). The PHQ-2 has been validated in both Spanish and Chinese with similar sensitivity and specificity among ethnically Latinx and Chinese populations as among non-Hispanic whites. [39][40][41][42][43] In our study, physicians were unaware of the PHQ-2 results at the time of visit as the measure was collected by research assistants during the telephone survey within 1 week after the index clinical visit.
Given that those with pre-existing depression could have already discussed their symptoms with their physician, or that the current elevated PHQ-2 score represented an improvement in symptom score, we limited our sample to those participants who were likely experiencing new symptoms. We excluded those with a prior history of depression listed on the electronic medical record (EMR) problem list before the index visit or who had a diagnosis of depression entered at a previous outpatient encounter (ICD-10 codes: F20.4, F31.3-F31.5, F32.x, F33.x, F34.1, F41.2, and F43.2).

Study sample
The LASI study identified 2359 potentially eligible participants; 662 (28%) were unreachable for the telephone interview. Among 1697 reached, 1181 agreed to participate (69.6% participation rate). Of these, 1171 participants (435 EP and 736 LEP) completed a PHQ-2 (99%). Two hundred twenty-one participants (19%) had current depressive symptoms. Of those, 118 participants (53%) had no history of depression and were included in this analysis.

Physician recognition of depressive symptoms
We defined physician recognition of depressive symptoms as a composite outcome, including any of the following: the physician documented a discussion of depressive symptoms in the clinical visit note, coded for depression (ICD-10 codes as above), initiated an antidepressant, or referred the patient to mental health services (including referral to social work for depression, psychiatry, or behavioral health services) at the index visit or at a follow-up visit within 30 days (a reasonable time frame to reassess for depressive symptoms 44 ). Data were obtained from the EMR and through chart reviews. We determined suicide risk assessment through chart review.

Race/ethnicity and English language proficiency
Latinx and Chinese ethnicity as well as gender were defined by self-report in the post-visit telephone survey. Participants were classified as having English proficiency or LEP based on a validated measure 45 of English language proficiency that includes patientreported preferred language for discussing health care and the patient's response to the U.S. Census question, ''How well do you speak English?'' 46 Response options include ''very well,'' ''well,'' ''not well, and ''not at all.'' Participants who spoke English ''very well'' or who spoke English ''well'' and preferred to discuss their health care in English were classified as having English Proficiency. 31 All others were classified as having LEP.
Prior year perceived mental health need Patient-reported perceived mental health need in the prior year was measured using a question adapted from the 2013 to 2014 California Health Interview Survey 47 and the 2002-2003 National Latino and Asian American Study 48 (CHIS and NLAAS; populationbased surveys): ''During the past 12 months, did you think you needed help for emotional or mental health problems, such as feeling sad, blue, anxious, or nervous?'' As has been defined in prior studies, an affirmative response indicated not only that participants had experienced mental health problems but also that they felt they needed help in addressing these problems. 21,49 Other covariates Additional demographics and participant characteristics included patient-reported age, education, health literacy, 50,51 and insurance type (private, Medicare, or MediCal). Participant clinical characteristics and health service utilization (from chart review and EMR queries) included Elixhauser comorbidities, 52,53 whether the participant had been in the practice for at least 1 year, the number of primary care practice visits in the prior year, the number of problems listed in the assessment and plan of the index visit, and whether the patient was seen by their own Primary Care Physician (PCP) or another physician at the index visit. We additionally categorized the index visits for patients with LEP as language concordant, discordant (either professionally interpreted or nonprofessionally interpreted), and partially concordant (see Table 1 footnote).

Statistical analysis
We computed the proportion of participants with current depressive symptoms with no prior history of depression, who had physician-recognized symptoms. We further examined each component of the composite outcome: documented discussion of depressive symptoms, initiation of an antidepressant, and referral for mental health treatment at the initial visit at the index visit or any visit in the clinical practice within 30 days. We examined bivariate relationships of covariates with physician recognition of depression symptoms, with a plan to conduct a multivariate logistic regression examining the association of race/ethnicity and language preference with our outcome.
Given that initial presentation of depression may include nonspecific mood symptoms, we then conducted a sensitivity analysis in which we included physician recognition of other mental health symptoms in our composite outcome. This included documentation of anxiety, bereavement, or trauma discussion or a mental health referral at the initial visit or a subsequent visit within 30 days of the index visit.
All analyses accounted for clustering of patients in the care of a physician. Stata version 14.2 was used for all calculations (StataCorp LP, College Station, TX). This study was approved by the University of California, San Francisco (UCSF) Institutional Review Board (IRB).

Sample characteristics and prevalence of current depressive symptoms
The 118 participants with current depressive symptoms and no history of depression tended to be older than 65 years (mean age 68), included majority women (70%), most had LEP (63%), had adequate health literacy (58%), were seen by their PCP during the index visit (72%), and were seen in languageconcordant visits in English or a non-English language (57%; Table 1).
Prior year perceived mental health need Seventy-one participants (61%) perceived a mental health need in the prior year. Chinese participants were more likely to perceive a mental health need than Latinx participants (71% vs. 40%; p < 0.001).

Physician recognition of depressive symptoms
In total, only eight unique participants (7% of those with PHQ-2 score ‡ 3) had a documented discussion of depression (Fig. 1). For these participants, physicians initiated an antidepressant for two participants (25%), both antidepressant and referral for two participants (25%) at the index visit, and placed a referral for two participants at a subsequent visit (25%). Two participants (25%) had a discussion about depressive symptoms, but no treatment was initiated. Out of the eight participants, four had negative suicide risk assessment (50%). Twenty-three participants (19%) with depressive symptoms had a follow-up visit with their primary care physician within 30 days of the index visit.
In the sensitivity analysis, including physician recognition of other mental health symptoms, a total of 13 (11%) unique participants discussed any mental health symptoms (8 depression, 3 anxiety, 1 grief, and 1 physical and emotional trauma). We found no significant difference in recognition and treatment We categorized the index visits for patients with LEP as language concordant (in English or non-English language), discordant-professionally interpreted, and discordant-not-professionally interpreted, and partially concordant. Visits were considered partially concordant if (1) physician reported ''none,'' ''poor,'' or ''fair'' skills in patient's preferred non-English language and patient reported speaking English ''Well,'' (2) physician reported ''good,'' ''very good,'' or ''excellent'' skills in patient's preferred non-English language, but failed language test in that language, or (3) physician reported ''good,'' ''very good,'' or ''excellent'' skills in patient's preferred non-English language, but they did not take the language test and had 3 or more patients give them an average rating of their skills as less than ''very good'' or ''excellent.'' AA, Associate of Arts Degree; LEP, limited English proficiency; PCP, Primary Care Physician; SE, standard error. initiation by race/ethnicity (Chinese vs. Latinx) or English proficiency (LEP vs. EP), although the number of participants with recognized symptoms was small. The number of patients with LEP with recognized symptoms was too small to evaluate for differences by language concordance or professional interpretation.

Discussion
Only a small number of patients with no history of depression had their depressive symptoms recognized and documented during a routine primary care visit. Even when we included physician recognition of other mental health symptoms (anxiety, grief, and trauma), in our definition of physician-recognized depressive symptoms, only 1 in 10 participants with depressive symptoms was recognized.
These findings are consistent with prior studies of poor provider recognition of current depressive symptoms in other primary care samples. [15][16][17] In another study using EMRs, only 12% of patients with moderate to severe depressive symptoms (based on a PHQ-8 score ‡ 10) were recognized over the course of a year. 15 In another prospective cohort study of primary care patients followed over 5 years, only 56% had their symptoms of major depression at baseline (diagnosed by psychiatric interview) recognized and treated over the study period. 17 However, many of the prior studies included few, if any, Asian and Latinx 21,54,55 patients. Poor recognition of depressive symptoms may contribute to the mental health disparities observed for ethnically Chinese and Latinx patients. 18,21,[26][27][28][29][30]56 In addition, older racial and ethnic minority patients, such as those included in our study, are at higher risk of poor recognition of depressive symptoms. Older adults may be more likely to have comorbidities causing competing demands in a primary care visit, be less likely to perceive a need for care, have limited knowledge and availability of affordable mental health services, and have difficulty arranging transportation. 21,54,55 The COVID-19 pandemic has made recognition and treatment of depressive symptoms increasingly critical. Due to the COVID-19 pandemic and subsequent shelter in place orders, there has been self-imposed isolation and loneliness, increased financial strain and unemployment rates, and a reduction in services, as well as the fear of becoming infected. This has placed individuals at increased risk for mental health symptoms. 12,14 Moreover, the pandemic has disproportionately affected racial and ethnic minorities in the United States. 57,58 To address the likely growth in service need for depression and anxiety, targeted efforts to recognize symptoms and engage racial and ethnic minorities in mental health treatment will be needed.
Barriers to recognition of depressive symptoms in primary care are multifactorial and include patient, physician, and system factors. Patients may present or focus on somatic complaints rather than classic mood symptoms, under-report or minimize depressive symptoms, or reject the diagnosis of depression due to stigma or cultural norms. [59][60][61][62] Physicians may be equally concerned about patient stigma regarding a mental health diagnosis and may avoid discussions about mental health symptoms or treatment. 63,64 Physician characteristics may influence comfort with discussing and addressing depressive symptoms, such as gender, racial, and ethnic concordance, years of practice, training background, implicit bias, setting, and familiarity with patients. [63][64][65][66] Patient-physician communication factors, such as language barriers, may further contribute to missing depression diagnoses. 30,49,63,67 In addition, physicians often have competing demands and time pressures during routine primary care visits, which may be further compounded for older adults with multiple comorbidities. 21,54 Systems factors include pressures on physicians to maintain productivity, limitations in mental health coverage for patients, limited access to specialty care, inadequate availability, and limited continuity of care, as well as few protocols and resources in place to systematically detect and manage patients with depression in many primary care settings. 68,69 The extent to which these factors played a role in the poor recognition of depressive symptoms in this sample of Chinese and Latinx primary care patients is unknown and should be further examined in future studies.
It is possible that physician ''recognition'' of depressive or other mental health symptoms may be more closely correlated with clinical action. In other words, physicians may recognize mental health symptoms, but not document these if they do not meet their threshold for clinical action. It is possible that some of the patients with elevated depressive symptoms on the PHQ-2, who did not receive a depression diagnosis or treatment/referral, were ''false positives'' for depression that does not meet a clinical threshold, or that due to the timing of our research study (asking patients about depressive symptoms within a week of the index visit), we may be misclassifying participants with less than 2 weeks of symptoms when seen by physicians. However, the high participant-perceived mental health need in this group suggests that many did have depression or another mental health condition at some point in the prior 12 months. Also, a positive response to the mental health need measure used in this study suggests that these patients wanted their mental health symptoms recognized and acknowledged, although it is possible that this mental health need had resolved before the index visit.
The small number of patients with physicianrecognized depressive symptoms limited our ability to investigate the individual contribution of factors that may affect physician recognition of depressive symptoms, such as language concordance, gender concordance, or continuity of care with physician. For participants with LEP, it remains unclear whether professional interpretation impacts recognition of depressive symptoms or if language concordance improves recognition.
Our study has limitations. The primary limitation is that we used a depression screening instrument outside of a clinical visit; patients may be more likely to disclose depressive symptoms in a research setting than a clinical setting, especially to bilingual, bicultural research assistants. In addition, we do not have a clinical assessment of depression as a gold standard or information about the severity of participants' depressive symptoms. It may be that those participants whose physicians did not recognize their symptoms by our definition only had minor depressive symptoms that may resolve over time without treatment; however, our inclusion of documentation of symptoms in any part of the clinical note rather than only a diagnosis casts a broad net to capture these patients as well. It is also possible that some of those who screened positive for depression did so due to other physical health conditions rather than depression, although the PHQ-2 is specific to mood and anhedonia rather than physical symptoms. 11 Physicians may also be accurately assessing the presence of another mental health condition or its co-existence with depressive symptoms, although including recognition of these symptoms in our definition of recognition of depressive symptoms only increased the percent of participants with recognized symptoms from 7% to 11%. We may also be including some individuals with a history of depression not documented in the EMR, and thus, despite our best efforts, the depressive symptoms may not be a new finding for physicians.

Conclusion
Primary care physicians may be failing to recognize and address current depressive symptoms in the large majority of their Chinese and Latinx patients (with and without language barriers) with no prior history of depression. As the COVID-19 pandemic has disproportionately impacted racial and ethnic minorities, and the number of individuals with mental health symptoms grows, this represents a potential missed opportunity and could lead to a further gap in treatment for Chinese and Latinx patients with depressive symptoms. While systematic depression screening may identify patients whose symptoms would otherwise go unrecognized, further training is also needed to increase provider detection and management of depression, particularly for racial and ethnic minority groups. It will be important to study barriers in the context of universal screening and identify additional supports needed to address depression treatment barriers for these vulnerable populations.

Acknowledgments
The statements presented are solely the responsibility of the author(s) and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute Ò (PCORI Ò ), its Board of Governors or Methodology Committee.

Author Disclosure Statement
The authors certify that they have no affiliation with or involvement in any organization or entity with any financial interest, or nonfinancial interest (such as personal or professional relationships, affiliations, knowledge, or beliefs) in the subject matter or materials discussed in this article.

Funding Information
Dr. Garcia received support from the University of California, Davis (UC Davis RCMAR/Latinx Aging Research Resource Center) under NIH/NIA Grant P30-AG043097; the Research in Implementation Science for Equity (RISE) Program, funded by NHLBI grant R25HL126146, the NHLBI K12HL138046, the Center for Aging in Diverse Communities (CADC) funded by NIH/NIH Grant P30-AG015272, and through an NIMHD K23MD015115; the content does not necessarily represent the official views of the NIA, NHLBI, NIMHD, or NIH. Dr. Garcia also wishes to thank SF BUILD (Building Infrastructure Leading to Diversity) for their support in the preparation of the article at a writing retreat. SF BUILD is funded by linked grants through the NIH Common Fund: UL1 GM118985; TL4 GM118986; and RL5GM118984. Research reported in this article was funded through a Patient-Centered Outcomes Research Institute Ò (PCORI Ò ) award (AD-1409-23627).