Mental Health Morbidities and Time to Cancer Diagnosis Among Adults With Colon Cancer in England

This cohort study investigates the association of preexisting mental health morbidities with time to diagnosis among adults with colon cancer in England.


Introduction
2][3] Mental health conditions are among the most frequent morbidities in Western populations, with 1 in 4 adults self-reporting mental health problems 4 and 1 in 8 adults (13%) presenting symptoms of anxiety or depression to primary care. 5MHM has been associated with advanced-stage cancer. 6][9][10] Diagnostic delays may occur when symptoms (eg, fatigue or abdominal pain, both possible colorectal cancer symptoms) are attributed to the preexisting MHM, which offers an alternative explanation. 8MHM may also be associated with delayed invasive investigations, such as colonoscopy, due to patient fear or anxiety 11 or competing priorities among patients and clinicians when managing complex clinical needs.Diagnosing cancer early, before it becomes a medical emergency, is paramount for improving survival.A 2022 population-based study 12 on colon cancer cohorts in 14 jurisdictions in Australia, Canada, New Zealand, Norway, Denmark, and the UK found that colon cancer diagnoses after emergency presentations occurred in between 23% and 36% of patients and were associated with 3-fold greater odds of 1-year mortality compared with nonemergency routes.A 4-month diagnostic delay of colorectal cancer may be associated with a 20% reduction of 10-year survival. 13,14The COVID-19 pandemic was associated with delays in cancer investigations, 13,15 with an estimated 17% increase in colorectal cancer deaths. 15It is estimated that after the pandemic, nearly 10 million people in England will have additional mental health needs, 16 meaning patients with MHM may experience doubly negative outcomes.This study focused on patients presenting with possible cancer symptoms to primary care given that approximately 85% of colon cancers are diagnosed after symptomatic presentations rather than screening. 17,18We aimed to examine variations in symptomatic presentations and subsequent diagnostic care by preexisting MHM status among patients with as-yet-undiagnosed colon cancer to characterize their odds of prolonged time to cancer diagnosis and emergency diagnosis.

Methods
This cohort study was approved by the UK Medicines and Healthcare Products Regulatory Agency (MHRA) Independent Scientific Advisory Committee under Section 251 (NHS Social Care Act 2006), which enables regulated access for research of anonymous data from the Clinical Practice Research Data link (CPRD), for which there is no requirement for informed consent.The study is based in part on data from the CPRD obtained under license from the UK MHRA.The data are provided by patients and collected by the NHS as part of their care and support.The study was performed in accordance with the Declaration of Helsinki and followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Study Population and Data Sources
We performed a cohort study of patients aged 30  Nonsymptomatic screening-detected cancers or emergency diagnoses without prior symptomatic primary care presentation were excluded.We also excluded rectal cancers and focused specifically on colon cancer given that diagnostic delays and emergency and advance-stage diagnoses occur more frequently in colon cancer (Figure 1). 19RD includes more than 670 UK general practices, is representative of the general population, and provides prospectively collected patient-level information on signs and symptoms, diagnoses, and tests. 20Cancer site and diagnosis date were obtained from the National Cancer Registry.Data linkage was performed by the trusted third party National Health Service Digital. 20Data analysis was performed in January 2021 to April 2022.

Study Variables
The main explanatory variables were MHM recorded in primary care before cancer diagnosis.
Validated code lists and definitions were obtained from the Cambridge Multimorbidity Score. 5ecifically, for anxiety or depression, read codes and 4 or more anxiolytic, hypnotic, or antidepressant prescriptions in last 12 months were considered. 5A combination of anxiety and depression is common, and treatments can overlap.We therefore examined anxiety, depression, or both grouped as 1 condition, in line with the Cambridge multimorbidity definition. 5For schizophrenia  or bipolar disorder, read codes or lithium prescription ever recorded were considered; alcohol problems, anorexia, and bulimia were defined based on read codes ever recorded. 5rther explanatory variables were red-flag colon cancer symptoms or signs, including rectal bleeding, change in bowel habit, or laboratory-confirmed anemia recorded in primary care before cancer; we also examined non-red-flag symptoms (eg, abdominal pain, constipation, diarrhea, weight loss, and fatigue). 21Change in bowel habit, rectal bleeding, and anemia are considered red-flag symptoms or signs warranting fast-track referral for investigations according to current guidelines. 21Relevant symptoms and signs (and related read codes or medcodes) were defined based on the literature and clinical expert revisions. 7,19,21Similar to previous work, 19 we classified symptoms as new onset if recorded for the first time during the 24 months before cancer diagnosis without records of the same symptom in the previous 3 to 5 years or as chronic symptoms if recorded during the 24 months and 3 to 5 years before cancer.Anemia was defined based on hemoglobin tests below sex-specific thresholds provided by CPRD.
Given that physical comorbidity may be associated with timeliness of diagnostic endoscopy, 19 we calculated the Charlson Comorbidity Index score using a validated algorithm, 22 identifying morbidity ICD-10 codes within HES inpatient and outpatient records.Additionally, we identified benign gastrointestinal (GI) conditions (ie, irritable bowel syndrome [IBS], diverticular disease [DD], and peptic ulcer) recorded in primary care before cancer given that these may be associated with MHM 16 and timely cancer diagnosis. 8We defined these conditions following Cambridge definitions; 5 for example, for IBS we considered read codes ever before cancer or 4 or more antispasmodic prescriptions in the last 12 months.Additionally, we included the number of primary care consultations for any reason during the year before cancer and sociodemographic characteristics (ie, sex, age, and socioeconomic deprivation using the Index of Multiple Deprivation 2015).

Primary Outcome Measures
The main outcomes were fast-track (also known as 2-week wait) referral for cancer investigations and emergency cancer diagnosis, defined according to the validated Routes to Diagnosis algorithm (eAppendix in the Supplement). 17,23Fast-track general practitioner (GP) referrals included patients referred urgently for suspected cancer to see a specialist within 2 weeks (introduced in England in 2000).Emergency cancer diagnoses included diagnoses after presentation to accident and emergency units or through GP emergency referral or emergency pathways for inpatients and outpatients. 17,23Additional outcomes included the overall symptomatic consultation to diagnosis interval, calculated similarly to what was done in previous work 19 for all patients with new-onset symptoms as the time in days from first new-onset relevant symptom recorded in CPRD during the 24 months before cancer diagnosis (index symptom) to cancer diagnosis (Figure 1).Secondary outcomes included use of colonoscopy or flexible sigmoidoscopy and interval from symptomatic consultation to endoscopy (days from index symptom to endoscopy).Endoscopy information was extracted from HES records in the 24 months before cancer and after a cancer-relevant symptom in primary care using OPCS 4.5 Standard Classification for National Health Services procedures codes (code list available upon request).A 2019 study 24 found 96% accuracy for HES data on investigations.Key variables and hypothesized associations among MHM, symptomatic presentations, and diagnostic routes are illustrated in the eFigure in the Supplement.

Statistical Analysis
We described sociodemographic characteristics, symptoms, physical comorbidity burden, number of GP consultations for any reason before cancer, endoscopy use, and diagnostic route by MHM status.The frequency distribution by MHM was compared using χ benign GI conditions, number of consultations, and sociodemographic characteristics.Given that general practices may vary in the use of endoscopy, 25 analyses accounted for patient clustering by practice and estimated robust standard errors.
We used quantile multivariable regression to examine variations in diagnostic intervals by MHM, accounting for symptoms, comorbidities, number of consultations, age, sex, and socioeconomic deprivation.This allowed us to compare diagnostic intervals by MHM and other patient characteristics by centile (50%, 25%, and 75%) of diagnostic intervals.Similar to previous research, we focused on the median (fiftieth centile) and seventy-fifth centile. 19cording to UK guidelines, fast-track referrals are warranted for patients with red-flag symptoms or signs. 21Thus, subanalyses focused on patients with red-flag symptoms or signs and examined the likelihood of fast-track referral by MHM status, accounting for patient characteristics.
Because fast-track referral is recommended for patients with colon cancer red-flag symptoms or signs aged 60 years or older, 21 we additionally examined this patient group.We used Stata statistical software version 16 (StataCorp) for statistical analyses.We used 2-sided tests and considered P < .05 as statistically significant.

Results
Among

Patient Characteristics and Route to Cancer Diagnosis by MHM Status
Patients with MHM were more frequently women, older, and more socially deprived than those without MHM.Additionally, patients with MHM less frequently had records of new-onset red-flag symptoms or signs (ie, rectal bleeding, change in bowel habit, or anemia).They had a higher burden of physical comorbidities, benign GI diagnosis (ie, IBS and DD) records and more primary care consultations; in contrast, they less frequently had records of endoscopy (Table 1  The symptomatic consultation to cancer diagnosis interval is presented for A, 3766 patients with symptomatic colon cancer and B, the subgroup of 2115 patients with new-onset red-flag symptoms or signs.The symptomatic consultation to endoscopy interval is presented for C, 2500 patients with symptomatic colon cancer who had an endoscopy and D, the subgroup of 1730 patients with red-flag symptoms or signs. Similarly, among all patients with colon cancer and any relevant symptom (not only red-flag symptoms), patients with MHM had higher odds of emergency diagnosis (adjusted OR = 1.38; 95% CI, 1.16-1.74;P = .001)(eTable 6 in the Supplement).

Discussion
In this cohort study, having an MHM was associated with 2-fold longer intervals before a colon cancer diagnosis.For individuals with new-onset red-flag symptoms, it took a median of 326 days (almost 1 year) from first symptomatic consultation to the diagnosis of cancer if they had a preexisting MHM

Interpretation in Context
[3]6 However, little was known about how MHM may be associated with the diagnostic process.Findings from 1 UK study and 2 small Dutch studies 7,8,26,27 suggested longer diagnostic intervals for patients with MHM.A 2022 Danish study 10 found that MHM was associated with a higher probability of cancer diagnosis through unplanned admissions, but whether this could be explained by differences in symptomatic presentations or timeliness of investigations was not explored.We found that patients with MHM and as-yet-undiagnosed colon cancer less frequently had records of typical cancer symptoms; furthermore, even when they had red-flag symptoms, they experienced disparities in prediagnostic care regarding use of and time to colonoscopy and routes to cancer diagnosis.
Little is known on how GI symptoms vary by MHM status in the general population without cancer.A study 28 on the association between MHM and IBS reported higher self-reported severity of GI symptoms in these patients, but the association between psychiatric and chronic GI conditions and symptoms is not well understood.MHM may be associated with reduced patient reporting or poorer interpretation and recording of potential cancer symptoms among doctors; symptoms, such as change in bowel habit, abdominal pain, or fatigue, may be considered outcomes associated with an underlying anxiety disorder 9 or adverse effects associated with psychotropic medications and attributed to benign conditions. 8Such mechanisms may partially explain why individuals with MHM less frequently had records of red-flag symptoms (including change in bowel habit) and more frequently had IBS records.IBS may provide an alternative explanation for cancer-delaying referrals in some patients, with longer symptom to diagnosis intervals and a higher emergency diagnosis risk, as observed in our study and previous studies. 8,27nsistent with earlier research, we found that high physical comorbidity burden was associated with prolonged diagnostic intervals, 7,8 lower odds of fast-track referrals 29 and endoscopy, 3,19 and higher odds of emergency presentation. 19While patients with MHM have been found to have higher comorbidity, 16 our study findings suggest that MHM and physical comorbidities are independently associated with timely cancer investigations.Some responsible mechanisms may overlap, while others may be specific to MHM.Patients with MHM may be reluctant to undergo colonoscopy due to fear or anxiety, 11 possibly associated with the lower odds of endoscopy observed in our study.Simultaneously, a high comorbidity burden may be associated with increased time before invasive investigations can be safely performed given that the risk of procedure-related complications needs to be appropriately managed. 30Our findings suggest that for patients with preexisting morbidities, including MHM and other comorbidities, competing demands may be associated with lower odds of timely investigations even in some patients with red-flag symptoms or signs, such as anemia or rectal bleeding. 8

Implications for Research and Practice
Our findings highlighted prolonged diagnostic intervals and possible missed opportunities for earlier diagnosis among patients with MHM presenting with red-flag cancer symptoms, which was associated with patient and health care factors.These may include difficulties in symptom appraisal and doctor-patient communication when making decisions on investigations, patient difficulties in adhering to recommendations (eg, bowel preparation for colonoscopy), 11 complexities in the informed consent process when patients have severe MHM with cognitive and emotional issues, 31 clinician cognitive bias, insufficient time and support during consultations, and limited availability of integrated pathways for patients with complex conditions. 6,8Some delays are difficult to avoid given that clinical presentation of MHM or medication-associated adverse effects may genuinely provide alternative explanations for cancer symptoms (eg, change in bowel habit); moreover, clinicians need to balance cancer risk with increased anxiety due to invasive procedures and cancer fear before referring patients with MHM for endoscopy.
To diagnose cancer earlier in the large number of patients with complex needs, improved support is necessary for patients and clinicians 1 given that multiple factors and suboptimal follow-up strategies may be associated with prolonged diagnostic process. 27Information technology based on electronic health records may help GPs to identify patients with complex conditions and plan resources and time allocation, for example, by involving specialist nurses before or after visits who are dedicated to patients with MHM.Similar approaches have been suggested in the management of physical multimorbidity. 32Information technology could also facilitate the provision of safety nets.Greater integration between primary and secondary care and wider use of disease-management programs coordinated by specialist nurses could enhance health care access. 1 Recently introduced Rapid Diagnostic Centres in the UK for patients with serious but nonspecific symptoms could also be beneficial in the case of diagnostic complexities due to MHM and multimorbidity.

JAMA Network Open | Public Health
Our study focused on patients diagnosed after symptomatic presentation given that they have worse outcomes than individuals detected via screening. 17Targeting patients who are symptomatic and facilitating screening are essential to diagnosing cancer earlier and improving outcomes. 33Future research based on interviews with patients and doctors examining the time between symptom onset and help-seeking (ie, the patient interval) and communication during clinical encounters may provide further insights into possible barriers to timely cancer investigations.

Limitations
Although we used validated algorithms and population-based data, encompassing prospectively recorded primary care, hospital, and cancer registration data, this study has several limitations.MHM was defined based on validated algorithms, including primary care diagnoses and symptom codes and prescriptions, obtaining estimates similar to those of previous studies 5 ; however, primary care records likely underestimate MHM because some individuals may not consult physicians or their conditions may not be recorded.Analyses by specific psychiatric diagnosis (eg, schizophrenia) and severity of MHM could not be performed due to small numbers and nonavailability of severity data.It is likely that delays in cancer diagnosis may be particularly pronounced in people with severe psychiatric conditions. 6,10We examined anxiety and depression grouped as 1 condition, in line with the Cambridge definition 5 ; however, future studies could examine these conditions separately given that their associated outcomes in GI-symptom management and cancer diagnosis may differ.Relying on routinely collected coded electronic records (without access to free-text notes) may have underestimated the number of patients who were symptomatic.
Physical comorbidity burden was measured using the validated Charlson Comorbidity Index based on hospital-recorded morbidities 22 without considering morbidities managed in primary care. 5 focusing on conditions severe enough to be recorded during hospital admissions before cancer, we accounted for morbidities likely associated with decision-making on invasive investigations, such as colonoscopy.Linked data were available to us up to 2015, and more recent data will be needed given persisting inequalities in diagnostic timeliness.
Given that our analyses included numerous explanatory and outcome variables, we limited the risk of overadjustment 6 by using different models for each outcome (fast-track referrals, emergency presentation, and diagnostic intervals) and considering unadjusted and adjusted models.Similar to previous research, 19 we focused on endoscopy use, which is well documented in HES. 24We did not examine the use of fecal calprotectin testing.A study 34 reported on its limited value for the diagnostic workup of patients suspected of significant colorectal disease.However, future research could investigate this further.We were not able to distinguish between primary care visits with a GP and those with another health care professional (eg, a primary care nurse).However, if red-flag symptoms are recorded during any primary care visit, appropriate consultations and testing should follow promptly.Our study found that this was not the case for a large proportion of patients.

Conclusions
This cohort study found that patients with MHM experienced large and prognostically consequential disparities in diagnostic care before a colon cancer diagnosis.Opportunities for earlier diagnosis may exist for a substantial subgroup of patients with colon cancer who also have anxiety, depression, or another MHM given that these patients were not investigated promptly for cancer despite presenting with red-flag symptoms.Our findings suggest that MHM may be independently associated with timely diagnosis of cancer, although physical comorbidity burden, older age, and social deprivation were also associated with lower odds of fast-track investigations.The prolonged

Figure 1 .
Figure 1.Study Sample, Data Sources, and Diagnostic Interval Definition

JAMA Network Open | Public Health Mental
to 99 years diagnosed with colon cancer (International Statistical Classification of Diseases and Related Health Problems, Tenth Revision [ICD-10] codes C18.1-C18.9) between 2011 and 2015 using National Cancer Registry records linked to Health Morbidities and Time to Cancer Diagnosis Among Adults With Colon Cancer in England JAMA Network Open.2022;5(10):e2238569. doi:10.1001/jamanetworkopen.2022.38569(Reprinted) October 31, 2022 2/15 Downloaded From: https://jamanetwork.com/ on 09/23/2023 primary care data (CPRD) and Hospital Episode Statistics (HES) outpatient and admitted patient data.We included 3766 patients with at least 1 colon cancer-relevant symptom recorded in primary care during the 24 months before cancer diagnosis (symptom list in eTable 1 in the Supplement).

JAMA Network Open | Public Health Mental
Colorectal cancer (CRC)-relevant symptoms included red-flag symptoms and signs (ie, rectal bleeding, change in bowel habit, and anemia) and nonspecific symptoms (eg, abdominal pain, constipation, diarrhea, weight loss, fatigue).CPRD indicates Clinical Practice Research Datalink; DCO, death certificate only; HES, Hospital Episode Statistics; NCRAS, National Cancer Registration and Analysis Service; RTD, route to diagnosis.Health Morbidities and Time to Cancer Diagnosis Among Adults With Colon Cancer in England ADiagnostic interval definitions

Table 1 .
). Emergency presentation was the most frequent diagnostic route for patients with MHM.IQR) symptomatic consultation to investigation interval was 118(29-453)days vs 72 (26-287) days (Figure2D).Details on the number of patients at risk per period are listed in eTable 3 in the Supplement).Among examined patient subgroups, only those with a Charlson Comorbidity Index score of 3 or higher had a similarly long median (IQR) interval (335 [110-571] days) (Table2).In Patient Characteristics and Route to Cancer Diagnosis Mental Health Morbidities and Time to Cancer Diagnosis Among Adults With Colon Cancer in England after accounting for covariables, we also found longer diagnostic intervals for individuals with MHM (fiftieth centile: adjusted interval, 224.4; 95% CI, 159.1-289.8;P=.003;seventy-fifthcentile:adjusted interval, 466.5; 95% CI, 413.4-519.6;P<.001)(Table2).

Table 2 .
Interval From First Consultation With Red-Flag Symptom or Sign to Cancer Diagnosis (n = 2115)Mental Health Morbidities and Time to Cancer Diagnosis Among Adults With Colon Cancer in England days (more than 4 months) for those without MHM.Having an MHM was associated with approximately 30% lower odds of prompt investigations after presentation with red-flag symptoms and 60% higher odds of cancer diagnosis through an emergency presentation, even after adjustment for other known risk factors.These findings suggest the possibility of missed opportunities for earlier diagnosis, especially but not exclusively among the large group of people with anxiety or depression.

Table 3 .
Likelihood of Fast-Track Referral Among Patients With Red-Flag Symptoms or Signs (n = 2115) Mental Health Morbidities and Time to Cancer Diagnosis Among Adults With Colon Cancer in England a Logistic regression ORs and 95% CIs are presented.bThe adjusted model includes all variables listed in the table.JAMA Network Open | Public Health JAMA Network Open.2022;5(10):e2238569. doi:10.1001/jamanetworkopen.2022.38569(Reprinted) October 31, 2022 9/15 Downloaded From: https://jamanetwork.com/ on 09/23/2023 Mental Health Morbidities and Time to Cancer Diagnosis Among Adults With Colon Cancer in England JAMA Network Open.2022;5(10):e2238569. doi:10.1001/jamanetworkopen.2022.38569(Reprinted) October 31, 2022 10/15 Downloaded From: https://jamanetwork.com/ on 09/23/2023 Routes to Diagnosis for Included Patients Who Were Symptomatic eFigure.Study Variables and Hypothesized Associations Between Mental Health Morbidity and Symptomatic Presentations and Route to Cancer Diagnosis eTable 2. Characteristics of 2115 Patients With Colon Cancer and Red-Flag Symptoms by Mental Health Morbidity (MHM) Status eTable 3. Patients at Risk for Diagnosis and Investigation by Mental Health Morbidity (MHM) per Time Point Before Cancer Diagnosis JAMA Network Open | Public Health Mental Health Morbidities and Time to Cancer Diagnosis Among Adults With Colon Cancer in England Likelihood of Being Diagnosed After Emergency Presentation (EP) in 2115 Patients With Red-Flag Symptoms eTable 5. Likelihood of Being Diagnosed After Expedited Referral in Patients Aged 60 y and Older With Red-Flag Symptoms eTable 6. Likelihood of Being Diagnosed After Emergency Presentation Among Patients With Colon Cancer Who Were Symptomatic JAMA Network Open.2022;5(10):e2238569. doi:10.1001/jamanetworkopen.2022.38569(Reprinted) October 31, 2022 14/15 Downloaded From: https://jamanetwork.com/ on 09/23/2023 eTable 4.