Why Are Mental Disorders Still Missed in Primary Care? (Journal Club Series)

Psychiatry Diagnoses Patterns, Not Lesions

In many areas of medicine, diagnosis is anchored by an observable lesion, an imaging finding, a laboratory abnormality, or a physiological measurement that narrows uncertainty. Psychiatry has always been different. Its central evidence is not usually a biomarker but a human pattern: what a person experiences, when it began, how it changed, what preceded it, what protects or worsens it, how it affects functioning, how it relates to family history, development, trauma, relationships, culture, sleep, substance use, and treatment response.

This makes psychiatric diagnosis unusually rich, but also unusually vulnerable. The diagnostic object is not a single symptom, nor even a checklist of symptoms, but a configuration unfolding across time. A panic attack, a period of insomnia, a depressive episode, an angry outburst, distractibility, avoidance, or fatigue may mean very different things depending on their temporal architecture and clinical context. The same symptom may be ordinary distress, a feature of major depression, part of post-traumatic adaptation, the depressive pole of bipolar disorder, an expression of ADHD-related dysregulation, a consequence of substance use, or a signal of medical illness.

The consequences of this difficulty are not theoretical. Mental disorders are highly prevalent in primary care: international primary-care data have estimated that roughly one-quarter of patients attending general medical settings meet criteria for at least one psychiatric disorder, and the World Health Organization has long emphasized the centrality of mental health problems in primary care populations (Goldberg & Lecrubier, 1995; World Health Organization, 1998, 2025). Yet detection remains strikingly limited. In a Canadian primary-care sample of 840 patients, non-detection or misdiagnosis reached 65.9% for major depressive disorder, 92.7% for bipolar disorder, 85.8% for panic disorder, 71.0% for generalized anxiety disorder, and 97.8% for social anxiety disorder (Vermani et al., 2011). A worldwide meta-analysis of anxiety detection in general practice reached a similar conclusion: without structured instruments, general practitioners detected anxiety disorders with a pooled sensitivity of only 30.5% (Olariu et al., 2015).

These numbers invite a tempting but misleading interpretation: that clinicians are simply failing to see what is in front of them. The evidence suggests something more interesting, and more humane. Psychiatric disorders are often missed because the relevant pattern is rarely fully in front of anyone during a brief clinical encounter.

A Psychiatric Diagnosis Depends on a Story

A psychiatric diagnosis depends on a story. Not a story in the sense of fiction or subjective embellishment, but a structured account of change over time. When did the symptoms begin? Were they episodic or chronic? Did they emerge before or after trauma, substance use, sleep disruption, hormonal change, grief, social adversity, or medical illness? Were there earlier developmental signs? Did family members show similar patterns? Did previous treatments help, worsen symptoms, or reveal hidden features of the disorder?

Bipolar disorder illustrates the point vividly. A person who presents in a depressive episode may look clinically indistinguishable from someone with unipolar depression unless the assessment includes lifetime mood history, treatment response, family history, periods of decreased need for sleep, activation, impulsivity, irritability, or functional change. Anchoring on the index depressive episode is understandable because it is what the clinician sees. But the diagnosis depends on what is not immediately visible: the longitudinal rhythm of mood (Hirschfeld et al., 2003; Vermani et al., 2011).

ADHD similarly depends on temporal reconstruction. Adult inattention can resemble anxiety, depression, trauma-related dissociation, sleep deprivation, substance use, or occupational overload. The diagnostic signal lies partly in developmental continuity: childhood onset, academic and interpersonal patterns, compensatory strategies, family history, and cross-situational impairment. PTSD also cannot be adequately understood from symptoms alone. Sleep disturbance, irritability, avoidance, concentration problems, numbing, and depression-like states become diagnostically meaningful when connected to trauma exposure, threat learning, re-experiencing, and functional adaptations over time (Friedman et al., 2011; Kessler et al., 2006).

This is the central paradox of psychiatric diagnosis in primary care: clinicians are often asked to recognize disorders that have developed over years, sometimes decades, using information gathered over minutes. The disorder is longitudinal; the encounter is cross-sectional. The condition is multidimensional; the visit is often organized around a single presenting complaint. The patient’s suffering may be diffuse; the clinical workflow requires a problem list, a diagnosis, a plan, and documentation before the next person is waiting.

Under those conditions, uncertainty is not a failure of reasoning. It is the natural result of missing context.

When Information Is Incomplete, Reasoning Becomes Fragile

Primary care encounters are often brief and problem-focused. Evidence links longer consultations with improved quality of care and more accurate identification of psychological problems (Hutton et al., 2007; Wilson & Childs, 2002). This matters because psychiatric information is often disclosed gradually.

Patients may begin with fatigue, headaches, abdominal symptoms, insomnia, pain, or stress rather than sadness, fear, trauma, obsessions, shame, or suicidal thinking. In the Canadian data, primary-care physicians frequently documented somatic correlates—headaches or backache, stress, gastrointestinal symptoms, insomnia—even when mood or anxiety disorders were not diagnosed (Vermani et al., 2011). That pattern is crucial. It suggests that clinicians often perceive distress signals, but the signals have not yet been assembled into a psychiatric configuration.

Human cognition enters the picture here, but not as a moral flaw. Clinical reasoning depends on mental shortcuts because medicine would be impossible without them. Dual-process models describe fast, intuitive, pattern-based reasoning and slower, more deliberate analytic reasoning; both are necessary, and both can fail under uncertainty (Croskerry, 2009; Norman & Eva, 2010). In a crowded primary-care day, fast reasoning is not laziness. It is adaptation to cognitive load.

The problem is that incomplete information makes heuristics fragile. Premature closure occurs when a plausible explanation is accepted before it has been adequately tested. Anchoring occurs when the first impression continues to dominate despite later contradictory information. Confirmation bias encourages the search for data that support an early hypothesis. Search satisficing stops the inquiry once one explanation has been found, even if comorbidity or a deeper differential remains unexplored (Croskerry, 2002, 2009; Saposnik et al., 2016). These biases are commonly implicated in diagnostic error, including primary-care diagnostic error, but they are best understood as vulnerabilities of normal cognition when a complex pattern must be inferred from partial evidence (Singh et al., 2013).

Symptoms Rarely Arrive Organized

In psychiatry, the risk is amplified because symptoms rarely arrive organized by DSM or ICD categories. Depression and anxiety frequently co-occur; physical and mental illness may be genuinely concurrent; bipolar disorder may first appear clinically as depression; PTSD may resemble depression or anxiety; adult ADHD may be misread as mood instability, anxiety, or poor motivation; and social anxiety may be mistaken for personality style rather than a treatable disorder (Hirschfeld et al., 2003; Katzelnick et al., 2001; Kessler et al., 2003; Meyer et al., 2020). Comorbidity is not an exception that complicates diagnosis after the fact. It is part of the diagnostic terrain from the beginning.

Somatic presentations make this terrain even more complex. Decades of work have described somatization as a major reason psychiatric illness remains hidden in general medical settings (Goldberg et al., 1988). This does not mean the symptoms are “not real.” It means psychological suffering may be communicated through the body, especially when emotional language is unavailable, stigmatized, culturally discouraged, or clinically uninvited. A patient with panic may present with chest tightness or shortness of breath. A patient with depression may present with pain, fatigue, insomnia, or gastrointestinal symptoms. A patient with trauma may present with diffuse physiological arousal and avoidance without naming trauma at all.

Clinicians may correctly recognize each symptom and still miss the disorder because the disorder is not contained in any one symptom. It is contained in the relationship among symptoms, history, meaning, timing, impairment, and context.

Seeing the Wrong Pattern—or No Pattern at All

Bias and stigma also operate through this same information problem. Diagnostic overshadowing occurs when symptoms are misattributed to an existing psychiatric, developmental, or medical label rather than investigated as part of a new or comorbid condition (Molloy, 2022; Molloy et al., 2023). Cultural and racial biases can distort which explanations feel plausible, which questions are asked, and which diagnoses become cognitively available. Evidence documents racial disparities in schizophrenia-spectrum diagnosis, lower screening rates for some linguistic groups, and measurable implicit associations linking Black individuals more strongly with psychosis-related concepts among clinicians in experimental settings (Ballard & Campinha-Bacote, 2025; Garcia et al., 2022; Westbrook et al., 2021). These are not merely interpersonal problems. They are failures of interpretive context: the clinician’s reconstruction of the patient’s pattern is shaped by what is asked, what is heard, what is assumed, and what remains unavailable.

A particularly revealing finding is that patient disclosure strongly predicts detection of depression and anxiety. When patients do not disclose symptoms, detection is unlikely; when symptoms are discussed, detection rises substantially (Marcus et al., 2011). This again points away from a simplistic model of missed diagnosis as poor recognition. Disclosure is part of the diagnostic data stream. It depends on time, trust, communication style, stigma, perceived safety, and whether the clinical encounter creates enough room for the patient’s experience to become visible.

Better Diagnosis Begins Before Clinical Reasoning

The most effective responses to missed psychiatric diagnosis share a common principle. They do not simply tell clinicians to “try harder.” They improve the information environment in which clinical reasoning occurs.

Structured tools, for example, increase the probability that relevant symptoms are elicited rather than left to chance. In anxiety disorders, standardized diagnostic aids raised sensitivity from 30.5% to 63.6% in general practice (Olariu et al., 2015). Brief tools such as the PHQ-2, PHQ-9, GAD-2, and GAD-7 show strong rule-out performance in primary-care contexts (Kroenke et al., 2001; Spitzer et al., 2006; Thombs et al., 2025). Structured interviews can substantially alter diagnostic conclusions compared with usual care, suggesting that relevant information often exists but is not systematically collected or synthesized (Bradford et al., 2024). Education, mental health literacy training, cognitive debiasing, diagnostic checklists, cultural competemility training, universal screening, and collaborative care may appear to be different interventions, but scientifically they converge on the same target: they make the diagnostic pattern more complete, more explicit, more equitable, and more available to reflection (Cheung et al., 2024; Garcia et al., 2022; National Academies of Sciences, Engineering, and Medicine, 2015).

Designing Around the Information Problem

This convergence has influenced Elyx’s design philosophy. One recurring observation in the evidence is that many diagnostic challenges arise before clinical reasoning formally begins. The issue is not only how clinicians interpret information, but whether the right information has been elicited, organized, contextualized, and made accessible in the first place. Elyx was therefore designed around a modest but scientifically important premise: collect richer, more structured, more longitudinal clinical information before the encounter, while preserving clinician autonomy and judgment.

This distinction matters. The goal is not to replace the clinician’s interpretive role, nor to reduce psychiatric diagnosis to automated scoring. The goal is to shift some of the most fragile parts of the diagnostic process upstream: developmental history, symptom chronology, previous treatments, family history, functioning, psychosocial context, comorbidity, risk factors, and the patient’s own language. In that sense, the relevant innovation is not simply technological. It is epistemological. It asks what kind of information psychiatry needs in order to reason well.

Psychiatry has never primarily been limited by a lack of compassionate or intelligent clinicians. Its challenge lies in the extraordinary complexity of understanding human beings across time. The missed diagnosis in primary care is not just a clinical quality problem; it is a window into the nature of psychiatric knowledge itself. Mental disorders are not always visible in the first complaint, the first impression, or the most available label. They are patterns that must be reconstructed from fragments.

Improving psychiatric diagnosis will likely depend less on replacing human judgment than on designing assessment processes that give judgment what it has always needed: richer context, better structure, more time, and a clearer view of the person’s trajectory.

In psychiatry, the diagnosis is rarely found in a moment; it is discovered in the pattern that connects moments across a life.

References

Ballard, T., & Campinha-Bacote, J. (2025). Cultural competemility training and use of a standardized assessment tool in reducing misdiagnosis of Black patients with schizophrenia spectrum disorders and psychotic disorders. Journal of the American Psychiatric Nurses Association, 31(3), 306–312. https://doi.org/10.1177/10783903241269046

Bradford, A., Meyer, A. N. D., Khan, S., Giardina, T. D., Singh, H., & Kunik, M. E. (2024). Diagnostic error in mental health: A review. BMJ Quality & Safety, 33(10), 663–675. https://doi.org/10.1136/bmjqs-2023-016996

Cheung, J., Chan, C. Y., & Cheng, H. Y. (2024). The effectiveness of interventions on improving the mental health literacy of health care professionals in general hospitals: A systematic review of randomized controlled trials. Journal of the American Psychiatric Nurses Association. https://doi.org/10.1177/10783903231194579

Croskerry, P. (2002). Achieving quality in clinical decision making: Cognitive strategies and detection of bias. Academic Emergency Medicine, 9(11), 1184–1204. https://doi.org/10.1111/j.1553-2712.2002.tb01574.x

Croskerry, P. (2009). A universal model of diagnostic reasoning. Academic Medicine, 84(8), 1022–1028. https://doi.org/10.1097/ACM.0b013e3181ace703

Friedman, M. J., Resick, P. A., Bryant, R. A., & Brewin, C. R. (2011). Considering PTSD for DSM-5. Depression and Anxiety, 28(9), 750–769. https://doi.org/10.1002/da.20767

Garcia, M. E., Ochoa-Frongia, L., Moise, N., Aguilera, A., Fernandez, A., & Lyles, C. R. (2022). Equitability of depression screening after implementation of general adult screening in primary care. JAMA Network Open, 5(8), e2227658. https://doi.org/10.1001/jamanetworkopen.2022.27658

Goldberg, D. P., & Lecrubier, Y. (1995). Form and frequency of mental disorders across centres. In T. B. Üstün & N. Sartorius (Eds.), Mental illness in general health care: An international study (pp. 323–334). John Wiley & Sons.

Goldberg, D. P., Bridges, K., Duncan-Jones, P., & Grayson, D. (1988). Detecting anxiety and depression in general medical settings. BMJ, 297(6653), 897–899. https://doi.org/10.1136/bmj.297.6653.897

Hallyburton, A., & Allison-Jones, L. (2022). Diagnostic overshadowing: An evolutionary concept analysis on the misattribution of physical symptoms to pre-existing psychological illnesses. International Journal of Mental Health Nursing, 32(1), 136–153. https://doi.org/10.1111/inm.13034

Hirschfeld, R. M. A. (2001). The comorbidity of major depression and anxiety disorders: Recognition and management in primary care. Primary Care Companion to The Journal of Clinical Psychiatry, 3(6), 244–254. https://doi.org/10.4088/PCC.v03n0609

Hirschfeld, R. M. A., Lewis, L., & Vornik, L. A. (2003). Perceptions and impact of bipolar disorder: How far have we really come? Results of the National Depressive and Manic-Depressive Association 2000 survey of individuals with bipolar disorder. Journal of Clinical Psychiatry, 64(2), 161–174. https://doi.org/10.4088/JCP.v64n0209

Hutton, C., Gunn, J., & Palmer, V. (2007). Do longer consultations improve the management of psychological problems in general practice? A systematic literature review. BMC Health Services Research, 7, Article 71. https://doi.org/10.1186/1472-6963-7-71

Katzelnick, D. J., Kobak, K. A., DeLeire, T., Henk, H. J., Greist, J. H., Davidson, J. R. T., Schneier, F. R., Stein, M. B., & Helstad, C. P. (2001). Impact of generalized social anxiety disorder in managed care. American Journal of Psychiatry, 158(12), 1999–2007. https://doi.org/10.1176/appi.ajp.158.12.1999

Kessler, R. C., Adler, L., Barkley, R., Biederman, J., Conners, C. K., Demler, O., Faraone, S. V., Greenhill, L. L., Howes, M. J., Secnik, K., Spencer, T., Üstün, T. B., Walters, E. E., & Zaslavsky, A. M. (2006). The prevalence and correlates of adult ADHD in the United States: Results from the National Comorbidity Survey Replication. American Journal of Psychiatry, 163(4), 716–723. https://doi.org/10.1176/ajp.2006.163.4.716

Kroenke, K., Spitzer, R. L., & Williams, J. B. W. (2001). The PHQ-9: Validity of a brief depression severity measure. Journal of General Internal Medicine, 16(9), 606–613. https://doi.org/10.1046/j.1525-1497.2001.016009606.x

Marcus, S. C., Olfson, M., Pincus, H. A., Shear, M. K., & Zarin, D. A. (2011). Detection of depression and anxiety disorders in primary care. Archives of Internal Medicine, 159(13), 1405–1411. https://doi.org/10.1001/archinte.159.13.1405

Meyer, F., Meyer, T. D., & Fuhr, K. (2020). Misdiagnosis of major depressive disorder in outpatient psychiatry: A structured interview study. Journal of Affective Disorders, 276, 671–678.

Molloy, R., Brand, G., Munro, I., & Pope, N. (2023). Seeing the complete picture: A systematic review of mental health consumer and health professional experiences of diagnostic overshadowing. Journal of Clinical Nursing, 32(9–10), 1662–1673. https://doi.org/10.1111/jocn.16151

National Academies of Sciences, Engineering, and Medicine. (2015). Improving diagnosis in health care. The National Academies Press. https://doi.org/10.17226/21794

Norman, G. R., & Eva, K. W. (2010). Diagnostic error and clinical reasoning. Medical Education, 44(1), 94–100. https://doi.org/10.1111/j.1365-2923.2009.03507.x

Olariu, E., Forero, C. G., Castro-Rodriguez, J. I., Rodrigo-Calvo, M. T., Álvarez, P., Martín-López, L. M., Sánchez-Totó, A., Adroher, N. D., Blasco-Cubedo, M. J., Vilagut, G., & Alonso, J. (2015). Detection of anxiety disorders in primary care: A meta-analysis of assisted and unassisted diagnoses. Depression and Anxiety, 32(7), 471–484. https://doi.org/10.1002/da.22360

Saposnik, G., Redelmeier, D., Ruff, C. C., & Tobler, P. N. (2016). Cognitive biases associated with medical decisions: A systematic review. BMC Medical Informatics and Decision Making, 16, Article 138. https://doi.org/10.1186/s12911-016-0377-1

Singh, H., Giardina, T. D., Meyer, A. N. D., Forjuoh, S. N., Reis, M. D., & Thomas, E. J. (2013). Types and origins of diagnostic errors in primary care settings. JAMA Internal Medicine, 173(6), 418–425. https://doi.org/10.1001/jamainternmed.2013.2777

Spitzer, R. L., Kroenke, K., Williams, J. B. W., & Löwe, B. (2006). A brief measure for assessing generalized anxiety disorder: The GAD-7. Archives of Internal Medicine, 166(10), 1092–1097. https://doi.org/10.1001/archinte.166.10.1092

Thombs, B. D., Corso, M., Wu, Y., Levis, B., Rice, D. B., Azar, M., & Benedetti, A. (2025). Screening tools for ruling out mood and anxiety disorders in adults in primary care: A systematic review. Journal of General Internal Medicine.

Tobon, A. L., Flores, J. M., Taylor, J. H., Johnson, I., Landeros-Weisenberger, A., & Bloch, M. H. (2021). Racial implicit associations in psychiatric diagnosis, treatment, and compliance expectations. Academic Psychiatry, 45(4), 533–534. https://doi.org/10.1007/s40596-020-01370-2

Vermani, M., Marcus, M., & Katzman, M. A. (2011). Rates of detection of mood and anxiety disorders in primary care: A descriptive, cross-sectional study. Primary Care Companion to CNS Disorders, 13(2), PCC.10m01013. https://doi.org/10.4088/PCC.10m01013

Wilson, A., & Childs, S. (2002). The relationship between consultation length, process and outcomes in general practice: A systematic review. British Journal of General Practice, 52(485), 1012–1020.

World Health Organization. (1998). Wellbeing measures in primary health care/the DepCare Project: Report on a WHO meeting, Stockholm, Sweden, 12–13 February 1998. World Health Organization.

World Health Organization. (2025, September 2). Over a billion people living with mental health conditions—services require urgent scale-up. World Health Organization.