Mind and Body, One Patient: How Mental Health Conditions Lead to Missed, Late, and Wrong Physical Diagnoses

Table of Contents

• Key Points
• Why This Research Matters
• What the Study Set Out to Investigate
• How the Researchers Found and Analysed the Evidence
• Key Findings: Diagnostic Inequalities Are Widespread
• What the Numbers Show: Cancer Diagnosis Delays, Study by Study
• What This Means for Patients and Health Services
• What This Research Could Not Prove
• Recommendations: What Needs to Happen Next
• Frequently Asked Questions
• Source Information

Key Points

• A review of 79 studies found that people with mental health conditions are at increased risk of delayed, missed, or incorrect physical illness diagnoses.
• Of 37 studies with reliable comparison groups, 29 showed a statistically significant increased risk of delayed physical health diagnosis.
• Diagnostic overshadowing, where physical symptoms are wrongly attributed to a mental health condition, is a recognised and dangerous form of diagnostic error.
• The burden should not fall solely on patients; health services and clinicians must also change how they work to reduce these inequalities.
• In one UK study of 2,115 patients with colon cancer signs, the diagnostic interval was 466 days for those with mental health conditions versus 365 days.

Why This Research Matters

Over the past twenty years, evidence has grown rapidly that people with mental health conditions die earlier than other people. In 2024, NHS England declared that people living with severe mental illness (SMI) — conditions such as schizophrenia, bipolar disorder, and severe depression — face "one of the greatest health equality gaps".

Globally, patients with SMI have a life expectancy 15–20 years shorter than the general population. But shorter life expectancy is not limited to severe illness. It is found across the whole spectrum of mental health conditions.

Most excess deaths in people with mental health conditions are caused by preventable physical illness. A 2019 review found that the risk of obesity, diabetes, and cardiovascular disease (heart disease and stroke) in this population is 1.4–2.0 times that of the general population.

Several factors are known to contribute to these physical health inequalities:

• Social and financial disadvantages
• Associated poorer lifestyles
• The impact of psychotropic medication (medication prescribed for mental health conditions) on physical health
• Suboptimal care for physical health problems

Inequalities in the diagnosis of physical health problems are also likely to play a role. Yet until this review, the evidence had never been systematically assessed.

Diagnostic inequalities are complex. They have multiple contributing factors and can happen at different stages of the diagnostic process. Diagnostic errors are defined as diagnoses that are missed, incorrect, or delayed. One well-known form is diagnostic overshadowing, where physical symptoms are wrongly attributed to a mental health condition. For example, chest pain might be dismissed as "anxiety" when it is actually a heart problem.

Importantly, diagnostic errors are not the only contributor to inequalities. Factors that happen before a patient even reaches a doctor matter too. Researchers use the concept of "total patient delay", which includes the time an individual takes to notice symptoms and seek medical care. People with mental health conditions may struggle to seek a diagnosis in a timely way because of the impact of their condition, financial burdens, healthcare models that do not meet their needs, and anticipation of stigma, poor-quality care, or not being taken seriously.

Both types of challenges matter, but they have different causes. While it is broadly accepted that people with mental health conditions are vulnerable to diagnostic inequalities, no overview of the research existed. This systematic review fills that gap.

What the Study Set Out to Investigate

The primary research question was direct: What evidence exists that individuals with mental health conditions are at risk of having their physical health problems undiagnosed, misdiagnosed, or diagnosed late?

The researchers also wanted to answer three secondary questions:

• Which physical and mental health conditions have been examined most frequently?
• What kinds of diagnostic problems have been assessed?
• At which point(s) of the diagnostic process do these problems occur?

The research aims were deliberately broad. The team wanted to provide a critical assessment of the current state of knowledge in this area of epidemiology (the study of how often diseases occur and why).

How the Researchers Found and Analysed the Evidence

The study was conducted and reported following PRISMA guidelines, an internationally recognised standard for reporting systematic reviews. Findings were synthesised narratively — in other words, described in a structured written summary rather than statistically combined. No meta-analysis (a statistical method that pools results from multiple studies) was conducted.

A peer researcher — someone with personal lived experience of mental health conditions — contributed to all stages of the review, including interpretation of the findings.

Because this review synthesises previously published content, ethical approval was not required.

How the search was conducted

Four major research databases were searched:

• MEDLINE (Ovid)
• Embase (Ovid)
• PsycINFO (EbscoHost)
• CINAHL (EbscoHost)

The initial search ran on 21 November 2022 and was updated on 18 September 2024. Additional studies were found through manual searches. The search strategy combined keywords and standardised index terms, developed using the PICO framework (a tool for framing health research questions) by the authors, including a medical librarian, in collaboration with experts in diagnostic error.

Studies written in English and published between 1 September 2002 and 18 September 2024 were included. Reviews, letters, editorials, comments, books, book chapters, case studies, and dissertations were excluded.

Titles and abstracts were screened independently by two reviewers. To be included, a study had to be a primary study (based on original data collection) using an established quantitative (numerical) design, providing information about physical health-related diagnostic inequalities in people with a mental health condition. Any healthcare setting in any country qualified.

Some conditions, such as dementia and delirium, sit at the intersection of mental and physical health. The authors classified them as mental health conditions because they involve significant psychological symptoms.

Studies were excluded if they were qualitative (non-numerical), if they dealt with diagnostic inequalities in mental health conditions themselves (for example, delayed diagnosis of bipolar disorder), or if they concerned people with intellectual or learning disabilities. Also excluded were studies of inequalities such as under-screening, undertreatment, or excess mortality that did not relate to diagnostic patterns, and studies of how common physical illness is in people with mental health conditions that did not examine diagnostic patterns.

How the data were analysed

Data were extracted by a single author and checked by an independent reviewer for accuracy. The extraction template was first piloted and refined on a sample of 10 studies. Risk of bias (systematic flaws that could skew results) was assessed using the Newcastle–Ottawa scale for non-randomised studies and the RoB 2 tool for randomised studies. These assessments were based solely on data relating to diagnostic inequalities, so they may not reflect the overall quality of each study.

Because the studies varied so much in design, definitions, measurements, conditions examined, and outcomes, no meta-analysis was conducted. Instead, the studies were grouped by the type of diagnostic problem they examined (studies suggestive of diagnostic error versus studies indicative of wider diagnostic inequalities) and, within those broad categories, by the physical and mental health conditions they targeted.

Key Findings: Diagnostic Inequalities Are Widespread

The search identified 18,966 articles from databases and 20 more from cross-referencing and manual searches. After removing duplicates, the abstracts of 13,428 papers were screened, and 13,066 were excluded. The remaining 362 papers were read in full, leading to the exclusion of 283. A total of 79 studies were eligible for inclusion.

Thirty-seven studies used a robust mental health comparator group (a group of people without mental health conditions, to allow fair comparison). Only these studies make it possible to measure the difference in risk of diagnostic inequality between people with and without mental health conditions. The review's main conclusions therefore focus on these 37 studies. The other 42 studies, which lacked such a comparison group, are summarised separately in the full paper's appendix.

The headline finding is stark: of the 37 studies with a robust comparator group, 29 found that having one or more mental health conditions is associated with a statistically significant increased risk of having a physical health problem delayed — as detected by a subsequent definitive test or finding.

Studies fell into two broad groups:

• Fifteen studies measured things suggestive of diagnostic error — missed, incorrect, or delayed diagnoses occurring after the patient presented to a health service. These designs eliminate late or non-presentation by patients as a possible cause.
• Twenty-three studies examined broader diagnostic inequalities, where the design could not distinguish between problems in the development of a diagnosis by health professionals and problems of patients presenting late or not at all.

In total, across the full set of 79 studies, 18 focused on diagnostic error (15 with a comparator group and 3 without), and 62 examined diagnostic disparities (23 with a comparator group and 39 without). One study (O'Rourke 2008) reported findings on both diagnostic error and diagnostic disparities, so it is counted in both groups.

A striking pattern emerged regarding the role of healthcare systems. Only a minority of studies (n=15) used a research design capable of isolating the specific role of health services and professionals in producing these inequalities. Of those, 14 found evidence that people with mental health conditions were at greater risk of diagnostic error (missed, wrong, or delayed diagnosis after seeing a clinician) than people without them. The remaining studies measured diagnostic endpoints only, meaning no conclusion could be drawn about the relative impact of patients' versus clinicians' behaviour.

What the Numbers Show: Cancer Diagnosis Delays, Study by Study

The visible portion of the review's evidence table focuses on cancer — the physical condition most frequently studied in relation to diagnostic delay. Here is what individual studies found. (A note on reading the numbers: a hazard ratio or odds ratio above 1 means higher risk or longer delay; below 1 can also mean longer delay when it measures the "hazard" of being diagnosed. Confidence intervals, shown in brackets, give the range the true value most likely falls within.)

• Colon cancer — Benitez Majano 2022 (UK): This study followed 2,115 patients with "red flag" signs of colon cancer, 308 of whom had a mental health condition (92% had mood disorders such as depression or anxiety). Median age was 75 (interquartile range 65–82). After accounting for other variables, the diagnostic interval (time from consultation to diagnosis) for people with mental health conditions was 466 days (95% CI 413–519) versus 365 days (95% CI 288.6–442.4) at the 75th percentile (p<0.001), and 224 days (95% CI 159–290) versus 126 days (95% CI 94.5–157.5) at the 50th percentile (p=0.003). Both differences were statistically significant.
• Colorectal cancer — Mounce 2017 (UK): In 4,512 patients with colorectal cancer, anxiety and depression were associated with longer diagnostic intervals from first symptomatic presentation to diagnosis: a 9-day diagnostic delay (95% CI 3–17), coefficient 0.11 (0.03, 0.20), p=0.007, adjusted for age and gender.
• Colorectal cancer — Van Hout 2011 (Netherlands): Among 197 patients (11 with mental health conditions, 5.6%), anxiety and depression were significantly associated with delay between first consultation with a general practitioner and referral to an endoscopy unit. The odds ratio (OR) was 3.87 (95% CI 1.13–13.30) in univariate analysis and remained significant in multivariate analysis at adjusted OR 3.97 (95% CI 1.14–13.85). This means people with anxiety or depression had nearly four times the odds of experiencing this delay.
• Colorectal cancer — Walter 2016 (UK): In a cohort of 2,507 (98.1% white, median age 65), people with anxiety or depression were diagnosed 0.8 times as quickly as those without (HR 0.8; 95% CI 0.71–0.90; p<0.001). The time from first symptom onset to diagnosis was also longer (HR 0.86; 95% CI 0.77–0.96; p<0.001). In plain terms, anxiety and depression meant a significantly longer overall diagnostic process.
• Breast cancer — Iglay 2017 (US): This study of 16,636 women aged 68 or older (3,961 with mental health conditions) found that women with comorbid anxiety or depression had an 11% increased risk of a diagnosis delay of 90 days or more, measured from first Medicare claim for breast symptoms to diagnosis (adjusted relative risk = 1.11; 95% CI 1.00, 1.23). No significant differences were found for other mental health clusters, or for any mental health condition when the threshold was a 60-day delay.
• Oesophageal cancer — O'Rourke 2008 (US): In 160 veterans (52 with mental health conditions), the median time from onset of alarm symptoms to diagnosis of oesophageal cancer was 90 days (IQR 20–162) for those with psychiatric illness versus 35 days (IQR 0–76) for those without, p=0.001. Multivariate analysis showed psychiatric illness in general (adjusted HR = 0.605; 95% CI 0.424–0.862) and depression specifically (adjusted HR = 0.622; 95% CI 0.425–0.910) predicted delayed diagnosis. Here, a hazard ratio below 1 indicates a lower "hazard" of being diagnosed — meaning a longer wait for diagnosis.
• Lung cancer — Iachina 2017 (Denmark): This study compared 508 people with depression to 27,234 without depression (controls) who were later diagnosed with non-small cell lung cancer. In contrast to the other studies, no difference was found in the duration of the diagnostic process between the two groups (adjusted HR = 0.99; 95% CI 0.90–1.09).

These findings show that the pattern is not universal — Iachina found no delay for lung cancer — but across the body of evidence, the overall trend is consistent. The conditions most strongly associated with the mortality gap, including cardiovascular disease, deserve particular attention, and the authors note that diagnostic inequalities related to cardiovascular problems should be a priority for future research.

What This Means for Patients and Health Services

The diagnostic inequalities identified in this review have potentially serious clinical consequences. A cancer that is diagnosed months later can be at a more advanced stage, harder to treat, and more likely to be fatal. The same logic applies to heart disease, diabetes, and other serious physical conditions.

A key insight for patients is this: the research shows that the problem is not simply about people with mental health conditions delaying their own care. Diagnostic errors — where clinicians miss, misinterpret, or take too long to reach a correct diagnosis even after the patient has sought help — are also common. Diagnostic overshadowing, where physical symptoms are written off as part of a mental health condition, is a recognised and dangerous form of this error.

Because professional and service-related factors make a clear contribution, the onus of behavioural change should not rest solely on patients. Health services, clinicians, and healthcare systems share responsibility for closing the gap.

For patients with mental health conditions and their families, this review offers both a warning and a tool. It is reasonable to be alert to the possibility of physical symptoms being overlooked, to prepare for appointments by describing symptoms clearly and specifically, and to ask clinicians directly: "Could this be a physical problem?" It is equally reasonable to expect clinicians to ask about physical symptoms in every consultation, including mental health appointments.

What This Research Could Not Prove

This review has important limitations, which the authors openly acknowledge:

• Cause and effect could not be determined. Most studies measured diagnostic "endpoints" only — the final outcome of delay or error. Because of their design, they could not show whether inequalities were caused more by patient behaviour, clinician behaviour, or system failures.
• Only a minority of studies examined the healthcare system's role. Just 15 of the studies were designed to isolate professional- and service-related factors, meaning the mechanisms behind the inequalities remain poorly understood.
• No meta-analysis was performed. The studies were too heterogeneous — they differed in design, definitions, measurements, conditions, and outcomes — to pool statistically. The findings are presented as a narrative synthesis instead.
• Risk of bias assessments were narrow. These were based only on the data relating to diagnostic inequalities, so they may not reflect the overall quality of each included study.
• Certain conditions were excluded. The review excluded diagnostic inequalities in mental health conditions themselves, and physical health diagnostic inequalities in people with intellectual or learning disabilities.
• Certain mental health conditions were under-represented. The authors note that personality disorders and eating disorders received little attention in the included studies.

Recommendations: What Needs to Happen Next

The authors call for a new wave of research that pinpoints which stage of the diagnostic process inequalities occur at. This is essential to understand the mechanisms at work and to design targeted improvement interventions.

Specifically, they recommend that future research should:

1. Consider the stage of the diagnostic process at which inequalities occur.
2. Focus on under-represented mental health conditions, particularly personality disorders and eating disorders.
3. Address diagnostic inequalities related to cardiovascular disease, which is the physical condition most strongly associated with the mortality gap.

For healthcare providers, the message is that measurement alone is not enough. Systems should be designed to catch physical illness early in people with mental health conditions. Practical steps might include structured physical health checks, decision support tools that counteract diagnostic overshadowing, and care pathways that coordinate mental and physical health care rather than treating them separately.

In the meantime, patients should know that this is a recognised problem receiving serious attention from the medical community. Being aware that the risk exists is itself a form of protection. So is persistence: if a physical symptom is not improving or is being attributed to a mental health condition, patients are entitled to ask for further investigation.

Frequently Asked Questions

What did this review find about physical health diagnoses in people with mental health conditions?

A review of 79 studies found that people with mental health conditions often face delayed, missed, or incorrect diagnoses for physical illnesses. Of 37 studies with reliable comparison groups, 29 showed a significantly increased risk of delayed diagnosis. This may help explain why this group dies 15–20 years younger than the general population.

What is diagnostic overshadowing?

Diagnostic overshadowing is when physical symptoms are wrongly attributed to a mental health condition. For example, chest pain might be dismissed as anxiety when it is actually a heart problem. This is a recognised and dangerous form of diagnostic error that can delay correct diagnosis and treatment of serious physical illnesses.

Does this mean patients with mental health conditions are to blame for late diagnoses?

No. The research shows the burden should not fall solely on patients. While some people may delay seeking care, studies also found diagnostic errors by clinicians after the patient had already sought help. Health services and clinicians must change how they work to close this gap.

What does a hazard ratio below 1 mean in these studies?

A hazard ratio below 1 can mean a longer wait for diagnosis. For example, in a study of 160 veterans with oesophageal cancer, an adjusted hazard ratio of 0.605 for psychiatric illness indicated a lower chance of being diagnosed promptly, meaning a longer time to diagnosis compared with those without psychiatric illness.

What can patients do if they are worried about a physical symptom?

It is reasonable to be alert to physical symptoms being overlooked. Prepare for appointments by describing symptoms clearly and specifically, and ask directly: 'Could this be a physical problem?' If a symptom is not improving or is being attributed to a mental health condition, you are entitled to ask for further investigation.

What are the limitations of this review?

The review could not prove cause and effect. Most studies measured final outcomes only, so they could not show whether patient, clinician, or system factors caused the inequalities. Only 15 studies examined the healthcare system's role. No meta-analysis was done because studies were too varied. Personality and eating disorders were under-represented.

If I have a mental health condition and my physical symptoms keep getting attributed to it, when should I seek a second opinion?

When a physical symptom is not improving, or is being attributed to a mental health condition, you are entitled to ask for further investigation. Diagnostic overshadowing — where physical symptoms are written off as part of a mental health condition — is a recognised and dangerous form of diagnostic error. Of 37 studies with robust comparison groups, 29 linked mental health conditions to significantly increased risk of delayed physical diagnosis. Prepare by describing symptoms clearly and asking directly: "Could this be a physical problem?" Diagnostic Detectives Network provides independent expert second opinions.

Source Information

Original article title: Diagnostic inequalities relating to physical healthcare among people with mental health conditions: a systematic review.

Authors: Liberati E, Kelly S, Price A, Richards N, Gibson J, Olsson A, Watkins S, Smith E, Cole S, Kuhn I, Martin G.

Journal: eClinicalMedicine, 2025, Volume 80, Article 103026. Published online by Elsevier Ltd.

DOI: 10.1016/j.eclinm.2024.103026

Funding: The study was funded by The Health Foundation's grant to the University of Cambridge for The Healthcare Improvement Studies (THIS) Institute.

The study was registered on PROSPERO (registration number CRD42022375892) and was conducted following PRISMA reporting guidelines.

This patient-friendly article is based on peer-reviewed research. It is intended for educational purposes and does not constitute medical advice. If you are concerned about physical symptoms, please speak with a healthcare professional.