Disparities in screening participation and stage at diagnosis for breast and bowel cancers among individuals with pre-existing mental health disorders: A population-based study.

Disparities in screening participation and stage at diagnosis for breast and bowel cancers among individuals with pre-existing mental health disorders: A population-based study

Abstract

People with pre-existing mental health disorders (PMHD) often have lower cancer screening uptake and later-stage diagnoses. It remains unclear whether these disparities persist within highly accessible universal healthcare systems without strict gatekeeping. We simultaneously investigated the associations between PMHD and both screening participation and stage at diagnosis of breast and bowel cancers within the Taiwan health service. Using nationwide National Health Insurance and Cancer Registry data covering over 99% of the population, we compared outcomes between individuals with and without PMHD. Propensity score matching was applied for the screening cohort. Adjusted odds ratios were estimated using logistic regression, with subgroup analyses for severe PMHD. Overall, PMHD was associated with higher screening participation (adjusted odds ratios [aOR]: breast = 1.04, bowel = 1.12) and reduced odds of metastatic disease at diagnosis (breast = 0.72, bowel = 0.93), although the magnitude of these associations was modest, particularly for bowel cancer. Conversely, individuals with severe PMHD had significantly lower odds of screening (breast = 0.81, bowel = 0.88) despite high outpatient use. Furthermore, while the severe group retained reduced odds of metastatic breast cancer at diagnosis (aOR = 0.79), they showed no staging advantage for bowel cancer. While Taiwan’s highly accessible universal healthcare system may contribute to a surveillance effect benefiting the general PMHD population, significant disparities persist for individuals with severe illnesses. Targeted interventions, procedural simplifications for self-administered tests, and multidisciplinary person-centred care models, may be needed to ensure health equity.

Introduction

Cancer and mental health disorders are both major global health concerns, showing significant co-morbidity. Individuals with pre-existing mental health disorders (PMHD) often experience significantly poorer cancer survival. While disparities in cancer outcomes are influenced by multiple factors such as treatment and comorbidities, stage at diagnosis is a critical determinant of prognosis [1]. Evidence indicates that PMHD is generally associated with advanced stage at diagnosis [2,3] and a prolonged interval from symptomatic consultation to diagnosis [4]. However, this association is not uniform. Some studies indicate no significant disparities, suggesting that outcomes may vary by mental health condition or healthcare system.

Screening is widely adopted as a primary approach to facilitate the earlier detection of cancers, but as a multi-step pathway, it may be less often undertaken by individuals with PMHD. While some studies show no significant disparities [5], evidence from other studies consistently indicates reduced uptake across diverse breast and bowel screening modalities [6–13]. These disparities stem from a combination of individual and systemic barriers. Individually, social, emotional, or cognitive disturbances can impair a patient’s ability to recognise physical symptoms or navigate complex healthcare pathways [2,5,6,14–16]. The impact of these difficulties is often exacerbated by the specific screening modality required, leading to issues such as poorer adherence to colonoscopy protocols and incomplete procedures [6,11,13,14,16]. Systemic barriers, including financial constraints and fragmented healthcare services, can lead to prolonged referral intervals and missed screening opportunities, compounding the risk of a late diagnosis [4,5,14–18]. Furthermore, the lack of trauma-informed care and negative experiences of compulsory treatment from mental health services impact trust in health services, resulting in lower cancer screening participation [5,19–22].

However, it remains unclear whether these disparities in screening participation and stage at diagnosis persist in healthcare systems with universal coverage and high accessibility. Most previous studies have reported from countries with healthcare gatekeeping or non-universal coverage healthcare systems, where systemic barriers may be the main challenges in cancer care among people with PMHD. In contrast, a system with high medical use theoretically provides a surveillance effect, providing frequent healthcare engagements for people with mental illness, which may increase both screening participation and opportunistic cancer detection [5,9,10]. Nevertheless, it is uncertain whether a high frequency of medical visits can compensate for barriers that impact people, such as trust in the system or a lack of trauma-informed care, particularly when screening or diagnostic procedures require supported and targeted outreach. Taiwan, with its National Health Insurance (NHI), offers a unique context to investigate the persistence of these disparities within a healthcare system characterised by universal coverage and high accessibility.

The Taiwanese NHI, implemented in 1995, has achieved 99% population enrolment and has service contracts with 92.44% of all health facilities [23]. Given its high healthcare accessibility and the absence of referral restrictions, the average per capita number of outpatient visits has reached 13.12 per year [23], providing a unique context with minimal systemic barriers to healthcare access. In addition, breast and bowel cancer screening programmes have been subsidised for high-risk populations for over two decades to address the high prevalence of these cancers in Taiwan. Moreover, population-based data on cancer registries, screening participation, and healthcare use have been comprehensively recorded in national databases for more than 20 years, enabling robust longitudinal analysis of the cancer care pathway for individuals with PMHD. Consequently, this study employs a nationwide population-based design to compare (1) cancer stage at diagnosis and (2) screening attendance between individuals with and without PMHD for breast and bowel cancers. Furthermore, we examine these disparities by focusing on individuals with severe PMHD to determine whether the severity of mental illness remains a persistent barrier to cancer care, even within a universal healthcare system characterised by high accessibility.

Materials and methods

Data sources and record linkage

This retrospective cohort study linked data from the National Health Insurance Research Database (NHIRD; 2016–2021) and Taiwan Cancer Registry (TCR; 2002–2021) using encrypted unique identifiers. NHIRD, covering 99.9% of the Taiwanese population, provides longitudinal claims data. It includes outpatient and emergency department visits, hospital admissions, and prescription records, which were used to identify pre-existing mental health disorders, comorbidities, and screening attendance. TCR provides histologically confirmed cancer diagnoses and detailed tumour staging information based on the American Joint Committee on Cancer (AJCC) system. The AJCC staging system 7th Edition was used for people newly diagnosed with cancer from 2016 to 2017, and the 8th Edition was used from 2018 to 2021.

Study design and population

To address the study aims, we established four distinct cohorts stratified by cancer type (breast and bowel) and study outcome (stage at diagnosis and screening attendance). These cohorts were organised under two primary analytical frameworks. Firstly, for the stage at diagnosis (S1–S3 Figs), individuals newly diagnosed with invasive breast cancer (International Classification of Diseases for Oncology, 3rd Edition [ICD-O-3]: C50) or bowel cancer (ICD-O-3: C18–C21) between 2016 and 2021 were included. Participants were excluded if they were (1) aged <18 years, (2) male (for breast cancer analysis), (3) diagnosed with carcinoma in situ (CIS), or (4) missing staging or demographic information (sex, income, or urbanisation). Secondly, for cancer screening attendance (S4–S6 Figs), the study population comprised individuals eligible for free breast or bowel cancer screening in 2019, including women aged 45–69 years for mammography and people aged 50–74 years for the Faecal Immunochemical Test (FIT). Individuals were excluded if they (1) were diagnosed with breast or bowel cancers prior to 2019, (2) had died during the observation period (2019–2021), (3) had a new breast or bowel cancer diagnosis during the observation, or (4) in the control group but diagnosed with a mental health illness during the study period, so ensuring a clean comparison. For the staging analysis, all eligible patients were included without matching to maximise statistical power, and potential confounders were adjusted through multivariable regressions. In contrast, the screening analysis compared individuals with PMHD against the general population. Given the potential substantial baseline disparities, a 1:5 propensity score matching using a greedy nearest-neighbour algorithm was applied to mitigate selection bias, matching individuals with PMHD to those without PMHD based on age, monthly income, and sex (for the bowel cohort only).

Definition of PMHD

PMHD was identified from diagnostic records over the 3-year period preceding the index date (1 January 2019 for the screening cohort and the date of cancer diagnosis for the staging cohort). Mental disorders were defined using codes from the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), and International Classification of Diseases, Tenth Revision (ICD-10) (complete list of codes provided in S1 Table). These codes were identified from inpatient, emergency, and outpatient claims. Disorders of organic or neurocognitive origin, such as dementia and delirium, were excluded. To ensure diagnostic validity and to exclude transient or tentative diagnoses, individuals were classified as having PMHD if they had at least one outpatient, emergency department visit, or inpatient admission with a primary diagnosis accompanied by the prescription of psychotropic medications, including antipsychotics, anxiolytics, hypnotics/sedatives, or antidepressants. Conversely, to ensure a distinct separation from those with PMHD, the comparison group consisted of individuals with no mental disorder diagnoses during the lookback and study observation period. Moreover, to investigate disparities in stage at diagnosis and screening participation for people with severe disorders, those people with a history of hospitalisation or emergency department visits for mental health conditions were identified from the PMHD group and categorised as having severe PMHD for sub-analyses.

Covariates

To account for potential confounding, statistical models were adjusted for personal and clinical characteristics. Regarding personal characteristics, age was calculated from the index date (1 January 2019 for screening; date of diagnosis for staging) and categorised into 5-year or 10-year age groups. Socioeconomic status was proxied by monthly insurance premium levels, which correlate with income tax brackets in Taiwan. Urbanisation was assessed using Liu’s seven-category system, ranging from 1 (most urbanised) to 7 (least urbanised) [24]. Clinical factors, including comorbidities, were measured using the Charlson Comorbidity Index (CCI) [25]. Scores were calculated based on diagnostic codes (ICD-9-CM/ICD-10-CM) recorded in inpatient and outpatient claims within the 2-year lookback period preceding the index date, and were classified into 0, 1, and ≥2 categories. Healthcare use was defined as the average annual number of outpatient visits calculated over the 3-year lookback period preceding the index date. This variable captures the individual’s longitudinal intensity of interaction with the healthcare system. Participants were categorised into three groups based on the distribution of these annual averages across the study population: < Q1 (low use), Q1–Q3 (moderate use), and >Q3 (high use). Cancer history, defined as a diagnosis of any cancer between 2002 and the index date, was obtained from the Taiwan Cancer Registry.

Outcome measures

Outcomes were assessed across two domains, including stage at diagnosis and screening attendance. Cancer stage was obtained from TCR based on AJCC TNM staging. In the primary analysis, cancer stage was categorised as non-metastatic disease (Stages I–III) versus (vs) metastatic disease (Stage IV), reflecting our aim to distinguish cancers without distant metastasis from those with distant metastasis at diagnosis. Although Stage III disease is generally considered locally advanced rather than early-stage disease, it remains non-metastatic. As a sensitivity analysis, cancer stage was additionally categorised as Stage I–II versus Stage III–IV to assess the robustness of the findings.

Regarding screening attendance, a 3-year observation period (2019–2021) was selected to fully capture compliance with the national biennial (2-year) screening guidelines, allowing for minor deviations in attendance timing. Breast cancer screening was defined as the uptake of at least one government-funded mammography, whereas bowel cancer screening was defined as the uptake of at least one FIT. While the bowel screening programme includes follow-up colonoscopy for positive cases, this study focused specifically on the initial uptake of FIT as the primary measure of screening accessibility and intent.

Statistical analysis

Continuous variables (means ± standard deviation [SD]) were compared using Student’s t-tests, and categorical variables (frequencies/percentages) using Chi-square tests. Missing data were excluded from the analysis. The staging analysis included all eligible newly diagnosed patients using multivariable adjustment, whereas the screening analysis used a 1:5 PSM procedure to mitigate selection bias. PSM was performed using a greedy nearest-neighbour algorithm without replacement, with a calliper width of 0.2 of the pooled standard deviation of the logit of the propensity score. Each patient with PMHD was matched to five controls (1:5 ratio). To assess covariate balance, standardised mean differences (SMDs) were calculated before and after matching (S2 Table), with an SMD < 0.1 indicating adequate balance [26,27]. Following these preparatory steps, multivariable logistic regression models were constructed to estimate adjusted odds ratios (aORs) and 95% confidence intervals (CIs). The variance inflation factor (VIF) was used to evaluate multicollinearity among covariates in the multivariable logistic regression models. VIF values of 2.5 or greater are considered indicative of considerable collinearity [28].

Separate models, adjusted for potential confounders, were developed to address the two objectives. For the cancer staging analysis, the odds of metastatic disease at diagnosis (Stage IV vs Stages I–III) were evaluated using multivariable logistic regression, with adjustments for age, sex (for bowel cancer), monthly income, urbanisation, CCI score, and history of cancer. Furthermore, a sensitivity analysis was conducted using the conventional Stages I–II versus Stages III–IV classification, with Stages III–IV representing locally advanced or metastatic disease, to evaluate the robustness of the study findings. Similarly, for the screening attendance analysis, despite the initial matching, models were further adjusted for age, sex (for bowel cancer), and monthly income to control for residual confounding, along with urbanisation, CCI score, and frequency of outpatient visits. To further understand the impact of severe disorders, all regression analyses were stratified to compare individuals with severe PMHD against the non-PMHD population.

All statistical analyses were performed using SAS software, version 9.4 (SAS Institute Inc., Cary, NC, USA), with statistical significance defined as p-value < 0.05. The study was approved by the Research Ethics Committee of China Medical University Hospital, Taiwan (No. CMUH114-REC2–023), and the requirement for informed consent was waived as this study solely used deidentified secondary data. Data were accessed for research purposes on 10 February 2025. The authors did not have access to information that could identify individual participants during or after data collection.

Results

Cancer stage at diagnosis

Between 2016 and 2021, a total of 77,417 women were newly diagnosed with breast cancer. Of these, 12,480 (16.1%) had PMHD, including 1,635 who were classified as having severe disorders. Women with PMHD were significantly older than those without. Notably, those with severe PMHD represented a particularly vulnerable subgroup with significantly higher proportions of low income and severe comorbidities (S3 Table). Despite these adverse characteristics and health conditions, after adjusting for personal and clinical covariates, women with PMHD were significantly less likely to be diagnosed at metastatic breast cancer (Stage IV: 7.3% vs 8.7%; aOR = 0.72, 95% CI = 0.67–0.78). Crucially, this inverse association with metastatic breast cancer diagnosis persisted even among those with severe PMHD (aOR = 0.79, 95% CI = 0.66–0.94) (Table 1).

Table 1Association between PMHD and stage at breast cancer diagnosis.

Variables Stage at breast cancer diagnosis　 Regression 1 Regression 2

Without vs with PMHD Without vs with severe PMHD

Stage I, II & III Stage IV p value (ROC = 0.597) (N = 66,572; ROC = 0.599)

n % n % OR 95% CI p-value VIF OR 95% CI p-value VIF

Total 70,880 91.6 6,537 8.4

PMHD <0.001 1.039 1.015

 Without (ref.) 59,307 91.3 5,630 8.7 1.00 1.00

 With PMHD 11,573 92.7 907 7.3 0.72 0.67 – 0.78 <0.001 – – –

 With severe PMHD 1,498 – 137 – – – – 0.79 0.66 – 0.94 0.009

Age at diagnosis (years old) <0.001 1.194 1.176

 <45 (ref.) 12,878 93.9 837 6.1 1.00 1.00

 45-54 20,528 93.0 1,543 7.0 1.16 1.07 – 1.27 <0.001 1.19 1.08 – 1.30 <0.001

 55-64 19,800 90.7 2,027 9.3 1.55 1.42 – 1.69 <0.001 1.59 1.45 – 1.74 <0.001

 65+ 17,674 89.2 2,130 10.8 1.72 1.57 – 1.88 <0.001 1.77 1.61 – 1.94 <0.001

Mean ± SD 56.6 ± 12.5 60.0 ± 13.0

 Monthly income (TWD) <0.001 1.056 1.058

 Under 22,000 (ref.) 15,593 88.9 1,945 11.1 1.00 1.00

 22,001-24,000 19,975 91.2 1,932 8.8 0.78 0.72 – 0.83 <0.001 0.77 0.72 – 0.83 <0.001

 24,001-40,100 15,268 92.2 1,300 7.9 0.75 0.69 – 0.80 <0.001 0.73 0.68 – 0.79 <0.001

 Above 40,101 20,044 93.7 1,360 6.4 0.59 0.55 – 0.64 <0.001 0.58 0.54 – 0.63 <0.001

Urbanisation <0.001 1.022 1.023

 Level 1 (ref.) 22,822 92.4 1,877 7.6 1.00 1.00

 Level 2 24,559 91.6 2,258 8.4 1.10 1.03 – 1.17 0.006 1.09 1.02 – 1.17 0.014

 Level 3 11,345 91.3 1,084 8.7 1.13 1.04 – 1.22 0.003 1.11 1.02 – 1.20 0.020

 Level 4 7,794 90.7 804 9.4 1.19 1.09 – 1.30 <0.001 1.19 1.08 – 1.30 <0.001

 Level 5 803 89.6 93 10.4 1.27 1.01 – 1.58 0.037 1.30 1.03 – 1.64 0.031

 Level 6 1,715 88.4 225 11.6 1.44 1.24 – 1.67 <0.001 1.40 1.19 – 1.65 <0.001

 Level 7 1,842 90.4 196 9.6 1.19 1.02 – 1.39 0.032 1.13 0.95 – 1.34 0.166

Charlson Comorbidity Score <0.001 1.157 1.143

 0 (ref.) 40,271 92.2 3,397 7.8 1.00 1.00

 1 18,110 91.5 1,677 8.5 1.02 0.96 – 1.08 0.597 1.04 0.97 – 1.11 0.281

 Above 2 12,499 89.5 1,463 10.5 1.17 1.10 – 1.26 <0.001 1.17 1.09 – 1.26 <0.001

Cancer history 0.003 1.019 1.019

 No (ref.) 66,880 91.5 6,226 8.5 1.00 1.00

 Yes 4,000 92.8 311 7.2 0.72 0.64 – 0.82 <0.001 0.72 0.63 – 0.82 <0.001

ROC: Receiver operating characteristic; OR: Odds ratio; CI: Confidence interval; PMHD: Pre-existing mental health disorder; TWD: New Taiwan Dollar; VIF: Variance inflation factor; SD: Standard deviation.

In parallel, regarding bowel cancer, 13,583 (17.2%) of the 79,142 newly diagnosed individuals had PMHD, with 2,639 classified as having severe disorders. Compared to those without disorders, individuals with PMHD were significantly older and female. Similar to the breast cancer stage cohort, individuals with severe PMHD represented the most vulnerable subgroup, being significantly older and having a higher comorbidity score (S4 Table). However, after controlling for personal and clinical covariates, individuals with general PMHD were significantly less likely to receive a diagnosis of metastatic bowel cancer (Stage IV: 24.4% vs 25.7%; aOR = 0.93, 95% CI = 0.89–0.98). In contrast to the breast cancer cohort, this inverse association was not observed among those with severe PMHD (aOR = 1.04, 95% CI = 0.95–1.14) (Table 2).

Table 2Association between PMHD and stage at bowel cancer diagnosis.

Variables 　Stage at bowel cancer diagnosis Regression 1 Regression 2

Without vs with PMHD Without vs with severe PMHD

Stage I, II & III Stage IV p value (ROC = 0.532) (N = 68,198; ROC = 0.532)

n % n % OR 95% CI p-value VIF OR 95% CI p-value VIF

Total 59,011 74.6 20,131 25.4

PMHD 0.003 1.035 1.020

 Without (ref.) 48,746 74.4 16,813 25.7 1.00 1

 With PMHD 10,265 75.6 3,318 24.4 0.93 0.89 – 0.98 0.002 – – –

 With severe PMHD 1,935 – 704 – – – – 1.04 0.95 – 1.14 0.402

Sex 0.546 1.010 1.006

 Female 25,520 74.5 8,755 25.5 1.00 1

 Male 33,491 74.7 11,376 25.4 0.98 0.95 – 1.02 0.304 1.00 0.96 – 1.03 0.782

Age at diagnosis (years old) 0.005 1.143 1.139

 <45 (ref.) 3,632 72.8 1,357 27.2 1.00 1

 45-54 8,174 74.0 2,879 26.1 0.94 0.87 – 1.02 0.126 0.94 0.87 – 1.02 0.128

 55-64 14,615 75.0 4,873 25.0 0.89 0.83 – 0.95 <0.001 0.89 0.83 – 0.96 0.002

 65+ 32,590 74.7 11,022 25.3 0.89 0.83 – 0.95 <0.001 0.87 0.81 – 0.94 <0.001

Mean ± SD 66.7 ± 13.5 66.8 ± 14.0

 Monthly income (TWD) <0.001 1.055 1.058

 Under 22,000 (ref.) 15,664 72.3 5,988 27.7 1.00 1

 22,001-24,000 20,064 74.8 6,753 25.2 0.87 0.84 – 0.91 <0.001 0.88 0.84 – 0.92 <0.001

 24,001-40,100 9,752 75.0 3,244 25.0 0.86 0.82 – 0.90 <0.001 0.85 0.81 – 0.90 <0.001

 Above 40,101 13,531 76.6 4,146 23.5 0.79 0.76 – 0.83 <0.001 0.79 0.75 – 0.83 <0.001

Urbanisation 0.009 1.024 1.026

 Level 1 (ref.) 16,196 75.5 5,245 24.5 1.00 1

 Level 2 18,550 74.1 6,500 26.0 1.07 1.03 – 1.12 <0.001 1.07 1.02 – 1.12 0.003

 Level 3 10,239 74.1 3,578 25.9 1.07 1.02 – 1.13 0.006 1.07 1.02 – 1.13 0.010

 Level 4 8,115 74.8 2,741 25.3 1.04 0.98 – 1.10 0.168 1.05 0.99 – 1.11 0.148

 Level 5 1,292 74.1 451 25.9 1.08 0.96 – 1.21 0.187 1.08 0.95 – 1.22 0.230

 Level 6 2,280 74.4 785 25.6 1.06 0.97 – 1.16 0.212 1.05 0.95 – 1.15 0.348

 Level 7 2,339 73.8 831 26.2 1.09 0.99 – 1.19 0.051 1.07 0.98 – 1.18 0.155

Charlson Comorbidity Score 0.137 1.120 1.115

 0 (ref.) 19,155 74.8 6,447 25.2 1.00 1

 1 16,894 74.1 5,910 25.9 1.05 1.01 – 1.09 0.027 1.06 1.01 – 1.10 0.014

 Above 2 22,962 74.7 7,774 25.3 1.03 0.99 – 1.07 0.212 1.02 0.98 – 1.07 0.314

Cancer history <0.001 1.010 1.011

 No (ref.) 54,067 74.2 18,756 25.8 1.00 1

 Yes 4,944 78.2 1,375 21.8 0.81 0.76 – 0.86 <0.001 0.80 0.75 – 0.86 <0.001

ROC: Receiver operating characteristic; OR: Odds ratio; CI: Confidence interval; PMHD: Pre-existing mental health disorder; TWD: New Taiwan Dollar; VIF: Variance inflation factor; SD: Standard deviation.

The sensitivity analysis using the Stages I–II versus Stages III–IV classification showed findings that were generally consistent with those of the primary analysis. For breast cancer, patients with PMHD had significantly lower odds of being diagnosed with Stages III–IV disease (aOR = 0.80; 95% CI = 0.76–0.83). Among women with severe PMHD, the adjusted odds ratio also indicated lower odds of Stages III–IV disease, although the association was not statistically significant (aOR = 0.98; 95% CI = 0.88–1.11). For bowel cancer, the results of the sensitivity analysis were consistent with those of the primary analysis (S5 Table).

Cancer screening attendance

In 2019, a total of 3,030,228 women in Taiwan were eligible for breast cancer screening. Data from 1,534,698 women were selected for further analysis after matching. Within this cohort, 255,783 women were identified as having PMHD, with 22,036 classified as having severe disorders. Women with severe PMHD represented a socio-economically disadvantaged and higher comorbidity score group. Notably, a significant disparity in healthcare use was observed, with 61.8% of women with severe PMHD falling into the highest quartile of outpatient visits, compared to only 19.7% of those without disorders (S6 Table).

Over a three-year period, 39.6% (n = 608,249) of eligible women engaged with breast cancer screening, and screening attendees exhibited a higher average number of outpatient visits than non-attendees (16.6 vs 10.3 visits/year) (Table 3). After adjusting for personal and clinical covariates, women with overall PMHD had a statistically significantly higher likelihood of participating in breast cancer screening (aOR = 1.04, 95% CI = 1.03–1.05). However, the findings reversed with disease severity. Even after accounting for the possible positive influence of frequent medical visits, women with severe PMHD had significantly lower odds of attending screening compared to the non-PMHD group (aOR = 0.81, 95% CI = 0.79–0.84) (Table 4).

Table 3Average outpatient visits.

Variables Without PMHD With PMHD With Severe PMHD

N Mean SD N Mean SD N Mean SD

Breast cancer screening

 Not attendance 797,911 10.3 12.1 128,538 24.3 17.5 12,430 28.0 21.0

 Attendance 481,004 16.6 13.1 127,245 29.0 17.6 9,606 33.9 20.7

Bowel Cancer screening

 Not attendance 1,461,736 10.5 12.8 229,204 24.6 18.4 24,155 27.4 21.6

 Attendance 773,894 18.8 14.6 217,922 30.7 19.2 18,164 35.4 22.2

PMHD: Pre-existing mental health disorder; SD: Standard deviation.

Table 4Association between PMHD and breast cancer screening attendance.

Variables Breast cancer screening Regression 1 Regression 2

Without vs with PMHD Without vs with severe PMHD

Not attendance Attendance p–value (ROC=0.678) (ROC=0.682)

n % n % OR 95% CI p–value VIF OR 95% CI p–value VIF

Total 9,26,449 60.4 6,08,249 39.6

PMHD <0.001 1.123 1.167

 Without (ref.) 7,97,911 86.1 4,81,004 79.1 1.00 1.00

 With PMHD 1,28,538 13.9 1,27,245 20.9 1.04 1.03 – 1.05 <0.001 – – –

 With severe PMHD 12,430 – 9,606 – 0.81 0.79 – 0.84 <0.001

Age (years) <0.001 1.062 1.058

 45–49 (ref.) 1,59,578 17.2 1,04,974 17.3 1.00 1.00

 50–54 1,82,925 19.7 1,23,189 20.3 0.94 0.93 – 0.95 <0.001 0.95 0.91 – 0.98 0.005

 55–59 2,03,161 21.9 1,36,367 22.4 0.85 0.84 – 0.86 <0.001 0.85 0.82 – 0.89 <0.001

 60–64 1,90,677 20.6 1,28,877 21.2 0.82 0.81 – 0.83 <0.001 0.82 0.79 – 0.85 <0.001

 64–69 1,90,108 20.5 1,14,842 18.9 0.68 0.67 – 0.69 <0.001 0.68 0.66 – 0.71 <0.001

 Mean ± SD 57.3 ± 7.0 57.2 ± 6.9 –

Monthly income (TWD) <0.001 1.036 1.033

 Under 23,100 2,97,777 32.1 1,21,029 19.9 1.00 1.00

 23,101–24,000 2,50,732 27.1 1,83,242 30.1 1.4 1.39 – 1.42 <0.001 1.42 1.37 – 1.46 <0.001

 24,001–40,100 1,93,818 20.9 1,46,652 24.1 1.49 1.48 – 1.51 <0.001 1.55 1.49 – 1.60 <0.001

 Above 40,100 1,84,122 19.9 1,57,326 25.9 1.73 1.71 – 1.75 <0.001 1.7 1.64 – 1.76 <0.001

Urbanisation <0.001 1.025 1.023

 Level 1 (ref.) 2,73,884 29.6 1,88,001 30.9 1.00 1

 Level 2 3,34,417 36.1 2,04,973 33.7 0.91 0.91 – 0.92 <0.001 0.91 0.88 – 0.94 <0.001

 Level 3 1,47,520 15.9 96,210 15.8 0.93 0.92 – 0.94 <0.001 0.91 0.88 – 0.94 <0.001

 Level 4 1,06,421 11.5 72,863 12.0 0.95 0.94 – 0.96 <0.001 0.91 0.88 – 0.95 <0.001

 Level 5 11,890 1.3 8,744 1.4 1.00 0.97 – 1.03 0.791 0.97 0.88 – 1.07 0.498

 Level 6 25,191 2.7 18,026 3.0 0.99 0.97 – 1.01 0.408 0.98 0.92 – 1.05 0.601

 Level 7 27,126 2.9 19,432 3.2 0.99 0.97 – 1.01 0.253 0.95 0.89 – 1.02 0.173

CCI <0.001 1.371 1.443

 0 (ref.) 6,34,238 68.5 3,23,573 53.2 1.00 1.00

 1 1,68,728 18.2 1,60,568 26.4 1.16 1.15 – 1.17 <0.001 1.19 1.16 – 1.23 <0.001

 Above 2 1,23,483 13.3 1,24,108 20.4 1.10 1.08 – 1.11 <0.001 1.06 1.03 – 1.1 0.001

Average annual outpatient visits <0.001 1.479 1.566

 <Q1 (Low) 3,35,552 36.2 61,285 10.1 1.00 1.00

 Q1~Q3 (Medium) 4,11,479 44.4 3,34,921 55.1 4.05 4.01 – 4.09 <0.001 4.11 3.97 – 4.26 <0.001

 >Q3 (High) 1,79,418 19.4 2,12,043 34.9 6.01 5.93 – 6.08 <0.001 6.24 5.97 – 6.51 <0.001

1. Q1 = 4, Q2 = 11, Q3 = 21

2. ROC: Receiver operating characteristic; OR: Odds ratio; CI: Confidence interval; PMHD: Pre-existing mental health disorder; TWD: New Taiwan Dollar; SD: Standard deviation.

In parallel, in 2019, a total of 5,453,628 individuals in Taiwan were eligible for bowel cancer screening, of whom 2,682,756 were included post-matching in the analyses. Among this subgroup, 447,126 individuals had a documented history of mental health disorders, with 42,319 classified as having severe PMHD. Similar to the breast cancer screening cohort, individuals with severe PMHD had higher proportions of lower monthly income and higher CCI scores (S7 Table). A distinct pattern was also observed, with 60.8% of individuals with severe PMHD falling into the highest quartile of outpatient visits, representing a rate nearly three times that of the non-PMHD group (22.2%).

Overall, 37% of study participants underwent FIT within the three-year study period. Similar to findings in breast cancer screening, attendees generally had a higher average number of outpatient visits than non-attendees (Table 3). Frequent outpatient visits exhibited a strong positive association with screening participation (aOR = 6.57 for >Q3 vs < Q1) (Table 5). Consistent with this pattern, individuals with overall PMHD, who generally used healthcare services more frequently, had a significantly higher likelihood of participating in bowel cancer screening compared to those with no disorders (aOR = 1.12, 95% CI = 1.11–1.12). However, this association was not observed in those with severe disorders. Individuals with severe PMHD were significantly less likely to undergo FIT screening (aOR = 0.88, 95% CI = 0.86–0.90) (Table 5).

Table 5Association between PMHD and bowel cancer screening attendance.

Variables Bowel cancer screening Regression 1 Regression 2

Without vs with PMHD Without vs with severe PMHD

Not attendance Attendance p-value (ROC = 0.705) (ROC = 0.714)

n % n % OR 95% CI p-value VIF OR 95% CI p-value VIF

Total 1,690,940 63.0 991,816 37.0

PMHD <0.001 1.112 1.150

 Without (ref.) 1,461,736 86.5 773,894 78.0 1.00 1.00

 With PMHD 229,204 13.6 217,922 22.0 1.12 1.11 – 1.12 <0.001 – – –

 With severe PMHD 24,155 – 18,164 – 0.88 0.86 – 0.90 <0.001

Sex <0.001 1.022 1.030

 Male (ref.) 823,635 48.7 389,913 39.3 1.00

 Female 867,305 51.3 601,903 60.7 1.36 1.35 – 1.37 <0.001 1.36 1.34 – 1.39 <0.001

Age (years) <0.001 1.078 1.065

 50-54 (ref.) 402,228 23.8 192,810 19.4 1.00

 55-59 404,329 23.9 227,297 22.9 1.07 1.06 – 1.08 <0.001 1.09 1.06 – 1.12 <0.001

 60-64 348,322 20.6 231,494 23.3 1.18 1.17 – 1.19 <0.001 1.20 1.17 – 1.23 <0.001

 64-69 309,033 18.3 228,447 23.0 1.21 1.20 – 1.22 <0.001 1.18 1.15 – 1.22 <0.001

 70-74 227,028 13.4 111,768 11.3 0.72 0.71 – 0.73 <0.001 0.72 0.69 – 0.74 <0.001

Mean ± SD 60.6 ± 6.9 61.1 ± 6.5

 Monthly income (TWD) <0.001 1.060 1.048

 Under 23,100 559,562 33.1 209,566 21.1 1.00

 23,101−24,000 434,457 25.7 320,763 32.3 1.42 1.41 – 1.43 <0.001 1.42 1.39 – 1.45 <0.001

 24,001-40,100 316,796 18.7 215,164 21.7 1.44 1.43 – 1.45 <0.001 1.43 1.39 – 1.46 <0.001

 Above 40,100 380,125 22.5 246,323 24.8 1.39 1.38 – 1.40 <0.001 1.43 1.39 – 1.46 <0.001

Urbanisation <0.001 1.027 1.024

 Level 1 (ref.) 487,444 28.8 275,026 27.7 1.00

 Level 2 585,778 34.6 329,048 33.2 1.03 1.02 – 1.04 <0.001 1.02 0.99 – 1.04 0.183

 Level 3 276,620 16.4 164,867 16.6 1.04 1.03 – 1.05 <0.001 1.03 1.00 – 1.06 0.025

 Level 4 205,105 12.1 131,116 13.2 1.09 1.08 – 1.10 <0.001 1.08 1.05 – 1.11 <0.001

 Level 5 26,029 1.5 17,994 1.8 1.13 1.11 – 1.16 <0.001 1.06 0.99 – 1.14 0.076

 Level 6 54,605 3.2 37,220 3.8 1.15 1.13 – 1.17 <0.001 1.11 1.06 – 1.17 <0.001

 Level 7 55,359 3.3 36,545 3.7 1.09 1.08 – 1.11 <0.001 1.07 1.02 – 1.12 0.005

Charlson Comorbidity Score <0.001 1.385 1.470

 0 (ref.) 1,193,506 70.6 509,578 51.4 1.00

 1 285,201 16.9 262,354 26.5 1.22 1.21 – 1.23 <0.001 1.17 1.15 – 1.20 <0.001

 Above 2 212,233 12.6 219,884 22.2 1.23 1.22 – 1.24 <0.001 1.16 1.13 – 1.19 <0.001

Average annual outpatient visits <0.001 1.424 1.599

 <Q1 (Low) 645,534 38.2 91,402 9.2 1.00

 Q1 ~ Q3 (Medium) 703,685 41.6 496,747 50.1 4.19 4.16 – 4.23 <0.001 4.56 4.44 – 4.69 <0.001

 >Q3 (High) 341,721 20.2 403,667 40.7 6.57 6.51 – 6.64 <0.001 7.51 7.27 – 7.75 <0.001

1. Q1 = 4, Q2 = 12, Q3 = 23

2. ROC: Receiver operating characteristic; OR: Odds ratio; CI: Confidence interval; PMHD: Pre-existing mental health disorder; TWD: New Taiwan Dollar; SD: Standard deviation.

Discussion

This population-based study reveals distinct cancer outcomes under Taiwan’s highly accessible universal healthcare system. Unlike disparities reported elsewhere, overall PMHD was associated with higher screening participation and reduced odds of metastatic stage diagnosis for both breast and bowel cancers. However, these advantages were largely absent for individuals with severe PMHD, despite their more frequent healthcare use. These findings suggest that while high healthcare accessibility may contribute to a potential surveillance effect, it may remain insufficient to mitigate the profound barriers associated with severe mental illness.

Potential surveillance effect

Our findings regarding higher screening uptake and reduced odds of metastatic disease at diagnosis among individuals with overall PMHD differ from previous literature. Reviews from other countries have synthesised findings to show that mental illnesses are associated with reduced screening participation and advanced stage at diagnosis for breast or bowel cancers [2,9,18]. Other studies attributed these disparities to cognitive disturbance [16,17] social isolation [14,16,17], transportation difficulties and costs [5,13,14,17], quality of the doctor–patient relationship [12,14,17], and fragmented healthcare services [6,14,15,17,18]. In contrast, Taiwan has no strict distinction between primary care and specialist care, and GP referrals to secondary care are not mandatory. This accessibility contributes to an average of 13.12 outpatient visits per person [23], double the Organisation for Economic Co-operation and Development average of 6.5 [29], thereby increasing exposure to healthcare services. We hypothesise that this distinct high frequency of contact may contribute to a potential surveillance effect for individuals with PMHD (Fig 1).

Within this accessible healthcare system, the specific advantage for individuals with PMHD is likely due to increased exposure to screening information through their frequent clinical contact. Our data indicate that individuals with PMHD attended outpatient visits significantly more frequently than those without PMHD. Screening attendees averaged 29–31 visits per year, compared with 17–19 visits among the non-PMHD group (Table 3). This high frequency may facilitate screening participation through two complementary mechanisms. Firstly, patients may encounter health promotion materials within the clinical environment, such as posters detailing breast and bowel cancer screening displayed in clinics and hospitals. Evidence shows that screening participation and knowledge consistently increased with the frequency of exposure to screening information [30]. Secondly, patients are subject to active systemic prompts, rather than relying solely on reminders from physicians during visits. Screening reminders are also automatically printed on prescriptions at each visit, which has been shown to effectively increase screening participation [31]. Consequently, frequent healthcare contact may increase opportunities to receive screening information through health education or recommendations from healthcare professionals. Notably, this systemic advantage may extend from screening participation to the diagnostic stage. The high frequency of healthcare visits may also increase the likelihood of timely recognition of alarm symptoms, such as palpable lumps or rectal bleeding, and subsequent referrals for examinations. These two potential mechanisms, including enhanced screening uptake and increased opportunities for symptom recognition, provide a possible hypothesis for the reduced odds of metastatic disease at diagnosis observed in the overall PMHD group.

Challenges for individuals with severe illnesses

However, contrasting findings for screening participation were found among people with a history of hospital admissions or emergency department visits for mental health disorders. Screening attendees with severe PMHD averaged over 33 visits per year, representing the highest among all groups (Table 3). Despite high medical use, these individuals were significantly less likely to attend breast or bowel cancer screening, which aligns with prior studies [14,31]. We therefore hypothesise that the advantage of frequent healthcare visits may not suffice to mitigate the barriers experienced by people with a diagnosis of severe mental disorders.

Individuals with severe disorders often have insufficient support from friends, family, or healthcare professionals, despite the demonstrated benefits of such support [6,15]. This may be particularly problematic for those who do not know what to expect during screening [15]. Needs for assistance in arranging cancer screening appointments have been revealed among this group [5,14,31]. In addition, individuals with a diagnosis of severe mental health disorders, their treating physicians, and other health practitioners, often focus on immediate medical concerns and complex mental health needs over the possibility of taking part in screening [6,12,13,15,18]. As a result, physicians may prioritise the management of acute mental health symptoms, leading to long-term preventive measures being overlooked or deprioritised during consultations [6,18]. For example, people with mental health disorders and red-flag symptoms are significantly less likely to receive prompt endoscopic investigations and experience diagnostic intervals that are more than twice as long as the general population [4].

Furthermore, symptoms of mental disorders, including depression, anxiety, and cognitive impairment, can hinder navigation of healthcare pathways and reduce motivation to attend preventive appointments [10,14,15,18]. This lack of engagement is particularly evident in bowel cancer screening. FIT-based screening requires home-based sample collection and return, which may evoke negative emotions such as disgust or shame and create difficulties in following sampling guidelines, thereby reducing participation [11,16,32,33]. Similarly, while breast cancer screening is professionally administered, individuals with mental health disorders have reported that participation is often inhibited by a lack of trust in unfamiliar staff, the anticipation of a traumatising procedure, and the fear of a cancer diagnosis [6,14,15].

Strengths and limitations

This study has several strengths. Firstly, the nationwide population-based design covers 99% of the Taiwanese population, minimising selection bias. Secondly, to our knowledge, this is the first study in Asia to simultaneously examine disparities at two critical phases of the cancer care pathway, including screening attendance and stage at diagnosis, within a universal healthcare system. This design allows for a more comprehensive understanding of potential inequality in cancer care among people with PMHD. Finally, the inclusion of both breast and bowel cancers enables a novel comparison between distinct screening modalities. While mammography is a professionally administered procedure performed entirely within clinical facilities, bowel cancer screening using FIT requires self-administration, involving home-based sample collection and return. By presenting these outcomes, our study provides insights into the association between mental illness and the differing engagement mechanisms screening requires.

However, several limitations should be noted. NHIRD relies on administrative claims data rather than medical charts. Therefore, we could not access physicians’ notes to verify the severity of mental symptoms or the exact reasons for non-attendance at screening. In addition, information on several potentially important factors was unavailable, including medication adherence, caregiver or family support, lifestyle factors such as smoking and alcohol consumption, sociodemographic factors such as educational attainment and marital status, psychiatric disease duration, and functional status. These factors may influence cancer screening behaviour and cancer stage at diagnosis and could therefore result in residual confounding. Moreover, NHIRD does not capture out-of-pocket healthcare services, potentially leading to an underestimation of overall screening participation. This missing information likely more often affects the non-PMHD group, as they may have greater access to unrecorded private screening. Consequently, our findings primarily reflect participation in government-funded screenings, suggesting a potentially greater dependence on government-subsidised screening services for the PMHD group.

Our definition of PMHD required a psychiatric diagnosis accompanied by psychotropic medication use and was intended to identify individuals with clinically recognised psychiatric conditions receiving pharmacological treatment using routinely available claims data. Nevertheless, this claims-based definition may not fully capture the spectrum of PMHD, particularly among individuals with psychiatric conditions who were not receiving pharmacological treatment. In addition, combining different psychiatric disorders into a single PMHD category may mask clinically meaningful differences in healthcare-seeking behaviours and barriers to cancer care. Further research is needed to explore whether cancer care disparities vary across specific mental health conditions and across different sex and ethnic groups which was not examined in this analysis.

The definition of severe PMHD relied on psychiatric hospitalisation or emergency department visits, which may serve as proxies for illness severity but do not directly measure psychiatric symptom severity or functional impairment. Therefore, some individuals with severe mental illnesses managed exclusively in outpatient settings may have been classified as having non-severe PMHD.

Excluding cases with missing stages or CIS could theoretically introduce selection bias. However, a supplementary analysis (S8 Table) comparing the excluded cohorts indicates minimal selection bias, suggesting that the observed surveillance effect may even be underestimated. Moreover, the relatively low discriminatory performance of the regression models, with ROC values of approximately 0.53–0.60, indicates that important determinants of cancer stage may not have been captured in the available datasets. In addition, although the associations between PMHD and advanced-stage cancer diagnosis were statistically significant, the magnitude of the observed associations was modest, particularly for bowel cancer (aOR = 0.93). Therefore, the clinical relevance of these findings should be interpreted with caution.

Finally, the three-year observation period (2019–2021) was selected to allow sufficient time to capture delayed or catch-up screening during the COVID-19 pandemic. However, this extended observation period may result in a higher cumulative proportion of screening participation than would be observed within a strict two-year interval and could therefore overestimate adherence to the recommended biennial screening schedule.

Clinical and research implications

Our findings suggest that an accessible universal healthcare system may mitigate health disparities for people with PMHD. The outcomes observed in cancer screening and staging for the overall PMHD population suggest that greater accessibility may help reduce disparities. For policymakers in countries with strict gatekeeping, Taiwan’s experience may provide support for policies that facilitate barrier-free access to medical professionals. However, universal healthcare coverage alone may not mitigate disparities among individuals with severe disorders. Our results demonstrate that frequent healthcare visits did not translate into better outcomes for those with severe mental disorders, highlighting the limitations of relying solely on a high-access system without more specific and targeted support.

Despite averaging over 33 outpatient visits annually, individuals with severe disorders remained significantly less likely to undergo screening. This highlights potential need for active interventions and procedural simplification. As potential future strategies, cancer patient navigators or specialist screening practitioners could provide cancer screening education and coordinate appointments, particularly for patients with functional impairments or insufficient social support. Moreover, people with lived experienced of mental health disorders (such as peer support workers or peer advocates) might also provide psychosocial support to facilitate engagement with screening. To address the logistical barriers of bowel cancer screening using FIT, future interventions could also evaluate the feasibility of “on-site” screening strategies. Collaborating with community care networks to facilitate local or postal sample return or providing facilities and assistance for patients to complete stool collections directly at the hospital during visits may reduce the challenges of bowel cancer screening. However, the effectiveness and acceptability of these approaches require evaluation in future research.

Moreover, the disparity in metastatic bowel cancer diagnosis and a higher proportion of stage unknown (S8 Table) for the severe disorders group highlights the necessity for targeted support to help these individuals. Further research is needed to explore the experiences and perspectives of people with severe PMHD, particularly their perceived barriers and enablers to physical health screening. Understanding these experiences may help inform the development of targeted interventions that are responsive to the needs of this population.

Conclusion

This nationwide study found that, within Taiwan’s universal healthcare system, people with PMHD generally had higher screening participation and lower odds of metastatic disease at diagnosis for breast and bowel cancers. However, those with severe PMHD had lower screening participation and remained at higher risk of metastatic bowel cancer at diagnosis. The universal healthcare system alone may be insufficient to mitigate disparities in cancer screening and stage at diagnosis among people with severe PMHD. Targeted interventions should therefore complement universal healthcare access, including simplified screening procedures and multidisciplinary, person-centred care, to promote greater equity in cancer screening participation and reduce disparities in stage at diagnosis.

Supporting information

Acknowledgments

We thank Tung-Han Tsai for running the analyses and Kuang-Hua Huang for obtaining the databases used in this study. We also appreciate the database provided by the Health and Welfare Data Science Centre of the Ministry of Health and Welfare, Taiwan.

Data Availability

The datasets (NHIRD and TCR) analysed during the current study are not publicly available due to strict privacy regulations governing personal health information in Taiwan. The data are maintained by the Health and Welfare Data Science Center (HWDC), Ministry of Health and Welfare, Taiwan. Access to the data can be requested through the HWDC data application system: https://dep.mohw.gov.tw/DOS/cp-5283-63826-113.html.

Funding Statement

The author(s) received no specific funding for this work.

References

1.Cunningham R, Sarfati D, Stanley J, Peterson D, Collings S. Cancer survival in the context of mental illness: a national cohort study. Gen Hosp Psychiatry. 2015;37(6):501–6. doi: 10.1016/j.genhosppsych.2015.06.003

2.Davis LE, Bogner E, Coburn NG, Hanna TP, Kurdyak P, Groome PA, et al. Stage at diagnosis and survival in patients with cancer and a pre-existing mental illness: a meta-analysis. J Epidemiol Community Health. 2020;74(1):84–94. doi: 10.1136/jech-2019-212311

3.Osterman E, Syriopoulou E, Martling A, Andersson TM-L, Nordenvall C. Mental illness and non-metastatic colorectal cancer treatment and survival, a nationwide study of almost 70,000 patients. Acta Oncol. 2025;64:585–94. doi: 10.2340/1651-226X.2025.42710

4.Benitez Majano S, Lyratzopoulos G, de Wit NJ, White B, Rachet B, Helsper C, et al. Mental Health Morbidities and Time to Cancer Diagnosis Among Adults With Colon Cancer in England. JAMA Netw Open. 2022;5(10):e2238569. doi: 10.1001/jamanetworkopen.2022.38569

5.Hwong AR, Vittinghoff E, Thomas M, Hermida R, Walkup J, Crystal S, et al. Breast Cancer Screening Rates Among Medicaid Beneficiaries With Schizophrenia. Psychiatr Serv. 2023;74(5):497–504. doi: 10.1176/appi.ps.20220163

6.Murphy KA, Stone EM, Presskreischer R, McGinty EE, Daumit GL, Pollack CE. Cancer Screening Among Adults With and Without Serious Mental Illness. Medical Care. 2021;59(4):327–33. doi: 10.1097/mlr.0000000000001499

7.Ross E, Maguire A, Mairs A, Hall C, Donnelly MJC, O’Reilly DPJ. Disparities in Breast Cancer Screening Uptake for Women With Mental Illness in the United Kingdom. Am J Prev Med. 2021;60(3):e123–30. doi: 10.1016/j.amepre.2020.09.010

8.Kerrison RS, Jones A, Peng J, Price G, Verne J, Barley EA, et al. Inequalities in cancer screening participation between adults with and without severe mental illness: results from a cross-sectional analysis of primary care data on English Screening Programmes. Br J Cancer. 2023;129(1):81–93. doi: 10.1038/s41416-023-02249-3

9.Jørgensen MD, Mikkelsen EM, Erichsen R, Thomsen MK. Mental illness and participation in colorectal cancer screening: a scoping review. Scand J Gastroenterol. 2022;57(10):1216–26. doi: 10.1080/00365521.2022.2073185

10.Kirkøen B, Berstad P, Hoff G, Bernklev T, Randel KR, Holme Ø, et al. Type and Severity of Mental Illness and Participation in Colorectal Cancer Screening. Am J Prev Med. 2023;64(1):76–85. doi: 10.1016/j.amepre.2022.08.011

11.Thomsen MK, Jørgensen MD, Pedersen L, Erichsen R, Sørensen HT, Mikkelsen EM. Mental disorders, participation, and trajectories in the Danish colorectal cancer programme: a population-based cohort study. Lancet Psychiatry. 2023;10(7):518–27. doi: 10.1016/S2215-0366(23)00179-7

12.Tuesley KM, Jordan SJ, Siskind DJ, Kendall BJ, Kisely S. Colorectal, cervical and prostate cancer screening in Australians with severe mental illness: Retrospective nation-wide cohort study. Aust N Z J Psychiatry. 2019;53(6):550–8. doi: 10.1177/0004867418814945

13.Seppänen A-V, Daniel F, Houzard S, Le Bihan C, Coldefy M, Gandré C. The double burden of severe mental illness and cancer: a population-based study on colorectal cancer care pathways from screening to end-of-life care. Epidemiol Psychiatr Sci. 2024;33:e27. doi: 10.1017/S2045796024000234

14.Tuschick E, Barker J, Giles EL, Jones S, Hogg J, Kanmodi KK, et al. Barriers and facilitators for people with severe mental illness accessing cancer screening: A systematic review. Psychooncology. 2024;33(1):e6274. doi: 10.1002/pon.6274

15.Linz S, Jerome-D’Emilia B. Barriers and Facilitators to Breast Cancer Screening for Women With Severe Mental Illness. J Am Psychiatr Nurses Assoc. 2024;30(3):576–89. doi: 10.1177/10783903221140600

16.Lin Y, Fan S, Chai W, Zheng N, Wang X, Wang Y, et al. Barriers and facilitators to population participation in colorectal cancer screening: an umbrella review. BMC Health Serv Res. 2026;26(1):288. doi: 10.1186/s12913-025-13879-z

17.Zubair NF, Mohammed Nawi A, Hassan MR, Daud A, Sarakarn P, Sutan R. Multilevel factors influencing colorectal cancer screening adherence: A systematic literature review. PLoS One. 2026;21(2):e0342184. doi: 10.1371/journal.pone.0342184

18.Aguilar‐Romero C, Benítez‐Hidalgo V, Ruiz‐Pérez I, Amaya‐Santos S, Pastor‐Moreno G. Breast cancer screening rates among patients with severe mental disorders (schizophrenia, bipolar disorder, and major depressive disorder). Systematic literature review and meta‐analysis. Psychooncology. 2025;34(10):e70314. doi: 10.1002/pon.70314

19.Jops P, Wolter A, Dodd R, Smith MA, Bavor C, Saunders T, et al. Toward equitable cervical screening: a qualitative study of under- and never-screened groups in Australia. SSM - Health Systems. 2026;7:100247. doi: 10.1016/j.ssmhs.2026.100247

20.Ruco A, Khalil A, Ledwos C, Tinmouth J, Kiran T, Lofters A. Exploring barriers and enablers to implementation of cancer screening among primary care professionals seeing marginalized patients. BMC Public Health. 2025;25(1):1578. doi: 10.1186/s12889-025-22835-9

21.Impelido ML, Brewer K, Burgess P, Curtis J, Currow D, Sara G. Age-specific differences in cervical cancer screening rates in women using mental health services in New South Wales, Australia. Aust N Z J Psychiatry. 2024;58(10):885–91. doi: 10.1177/00048674231217415

22.Mkuu RS, Staras SA, Szurek SM, D’Ingeo D, Gerend MA, Goede DL, et al. Clinicians’ perceptions of barriers to cervical cancer screening for women living with behavioral health conditions: a focus group study. BMC Cancer. 2022;22(1):252. doi: 10.1186/s12885-022-09350-5

23.National Health Insurance Administration M of H and W. National Health Insurance Annual Report 2022-2023 (bilingual). National Health Insurance Administration Ministry of Health and Welfare. 2022. https://www.nhi.gov.tw/en/cp-11-434ff-59-2.html

24.Liu CY, Hung YT, Chuang YL, Chen YJ, Weng WS, Liu JS. Incorporating development stratification of Taiwan townships into sampling design of large-scale health interview survey [In Chinese]. Journal of Health Management. 2006;4(1):1–22. doi: 10.29805/JHM.200606.0001

25.Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi J-C, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43(11):1130–9. doi: 10.1097/01.mlr.0000182534.19832.83

26.Zhang Z, Kim HJ, Lonjon G, Zhu Y, written on behalf of AME Big-Data Clinical Trial Collaborative Group. Balance diagnostics after propensity score matching. Ann Transl Med. 2019;7(1):16. doi: 10.21037/atm.2018.12.10

27.Wan F. Propensity score matching: should we use it in designing observational studies?. BMC Med Res Methodol. 2025;25(1):25. doi: 10.1186/s12874-025-02481-w

28.Johnston R, Jones K, Manley D. Confounding and collinearity in regression analysis: a cautionary tale and an alternative procedure, illustrated by studies of British voting behaviour. Qual Quant. 2018;52(4):1957–76. doi: 10.1007/s11135-017-0584-6

29.OECD. Health at a glance 2025. doi: 10.1787/8f9e3f98-en
2025.

30.Baskar S, Schoeneich R, Baskar A, Grewal US. Leveraging Patient Education to Amplify Colorectal Cancer Screening in the United States: Strategies and Implications. J Cancer Educ. 2025;40(3):321–8. doi: 10.1007/s13187-024-02482-1

31.Larsen MB, Hedelund M, Flander L, Andersen B. The impact of pre-notifications and reminders on participation in colorectal cancer screening - A randomised controlled trial. Prev Med. 2022;164:107229. doi: 10.1016/j.ypmed.2022.107229

32.Reynolds LM, Bissett IP, Consedine NS. Emotional predictors of bowel screening: the avoidance-promoting role of fear, embarrassment, and disgust. BMC Cancer. 2018;18(1):518. doi: 10.1186/s12885-018-4423-5

33.Scaglioni G, Cavazza N. Emotional Barriers to Bowel Screening in Italy: Scale psychometric properties and effects on screening attendance. Psychooncology. 2022;31(1):78–85. doi: 10.1002/pon.5781

Associated Data

Supplementary Materials

Data Availability Statement

The datasets (NHIRD and TCR) analysed during the current study are not publicly available due to strict privacy regulations governing personal health information in Taiwan. The data are maintained by the Health and Welfare Data Science Center (HWDC), Ministry of Health and Welfare, Taiwan. Access to the data can be requested through the HWDC data application system: https://dep.mohw.gov.tw/DOS/cp-5283-63826-113.html.