Mental health, health status, and recurrent healthcare utilization in patients with non-obstructive coronary artery disease

Participants and procedure

Data came from the TWeeSteden mIld STenosis (TWIST) study, a single-center prospective cohort on psychosocial risk factors for NOCAD progression (25). Between 2009 and 2013, patients were recruited at ETZ Hospital (Tilburg, NL). Inclusion required a CT calcium score >0 without subsequent CAG referral, or CAG-detected vascular irregularities/mild stenosis; exclusions were normal coronaries, obstructive IHD, prior obstructive events, or insufficient Dutch proficiency. Of 6490 screened, 883 were eligible and invited; 546 (61%) consented and were followed. Questionnaires were completed at baseline, 12, and 24 months; hospital records were reviewed at baseline and ∼10 years. The study was approved by the METC Brabant (NL22258.008.08, amendment ETZ: L1160.2020/LP.2008.227) and conducted in accordance with the Declaration of Helsinki

Recurrent cardiac health care utilization [HCU]

Recurrent cardiac HCU was retrieved from hospital records (2021/2022)(16, 26). Data included ED visits for cardiac reasons (event and date) and recurrent diagnostic tests (CT, CAG, SPECT, or MRI) since baseline. Total HCU was calculated as the sum of ED visits and retests; events in close succession (e.g., ED visit followed by CAG) were counted separately. Patients without events were censored at the final screening date (Feb 3, 2022); those lost to follow-up were censored 12 months after last contact (16). Time was described as average time between events (or censoring) and maximal study time (baseline CAG/CT to final/censored date).

Mental health and health status

Validated questionnaires were used to assess mental health and health status. Anxiety and depressive symptoms were measured with the Hospital Anxiety and Depression Scale (HADS; 14 items, subscales 0–21 for anxiety [HADS-A] and depression [HADS-D])(27). Type D personality was assessed with the DS14 (two 7-item subscales: social inhibition [SI] and negative affectivity [NA], range 0–28)(28, 29). Type D was modeled using standardized SI and NA scores plus their interaction (29).

Health status was assessed with the Fatigue Assessment Scale (FAS; 10 items, higher scores = more fatigue) (30), the 12-item Short Form (SF-12; physical [PCS] and mental [MCS] summary scores, higher = better health) (31), and a modified 19-item Seattle Angina Questionnaire (mSAQ) adapted for broader cardiac symptoms (25, 32). The mSAQ measured physical limitation, angina stability, angina frequency, treatment satisfaction, and quality of life(/disease perception), with higher scores indicating better health status (33).

Covariates and descriptives

At baseline, patients completed a questionnaire on sociodemographics (sex, age, partner status, education, employment status) and lifestyle factors (BMI, obesity, smoking, physical activity, alcohol use). Cardiac risk factors and disease severity were obtained from hospital records: diabetes, hypertension, dyslipidemia, diagnostic exam at inclusion (CAG versus CT scan), visible wall irregularities in three vessels (LAD, CX, RCA), chest pain in the past month, and classification of patients as ANOCA (chest pain symptoms), INOCA (chest pain and detected ischemia), or other (familial/high cardiovascular risk)(34).

Progression of IHD; Major Adverse Cardiac Events (MACE)

Because recurrent HCU may indicate IHD progression, patients were classified by presence or absence of MACE, used as a proxy for disease progression. MACE, derived from hospital records, included MI, PCI, CABG, heart failure, and cardiovascular mortality (16). Events were recoded into progression (MACE) versus no progression (No MACE). Noncardiac mortality was treated as a censored event and could occur in any subgroup.

Statistical analyses

For descriptive purposes, three HCU groups were defined: patients who had no HCU since baseline, those with a single HCU (either an ED visit or retest), or all patients with ≥2 HCU events (multiple HCU). The prevalence of events per HCU group was described as the occurrence of recurrent ED visits, retesting, total HCU, MACE (including cardiac and noncardiac mortality), and cardiac and noncardiac mortality. Time since baseline CAG/CT scan, maximum study time, and average time to any event or censoring were also described. Group differences in sociodemographic variables, lifestyle, risk factors, disease severity, mental health, and health status were assessed using Chi-square tests and One-way ANOVA with post-hoc analyses. For variables with ≥2 categories (e.g., alcohol use, NOCAD subgroups), linear-by-linear associations were reported.

To identify covariate predictors of HCU regularized regression was performed in R (glmnet). A 10-fold cross-validation determined the optimal alpha (ridge, LASSO, or elastic net) regression. Missing data were imputed using mice with predictive mean matching in 10 datasets. Covariates consistently associated with HCU across imputations were included in the final prediction models.

The main outcome was total HCU (count). Due to excess zeros and overdispersion, negative binomial regression with log link was used instead of Poisson regression (35). Mental health and health status variables were tested in separate models, except for Type D traits and their interaction, which were examined together, adjusted for age and sex. Additional covariates were added based on predictors identified through regularized regression, and subsequently for IHD progression dichotomized as no MACE vs. any MACE. Analyses used pooled estimates from five imputed datasets, with factors coded in descending order (reference = 0). Maximum iterations were set to 50.

Explorative analyses were conducted by stratifying the prediction model by sex and stratified for MACE (no versus any) for each mental health and health status factor separately, adjusted for all other covariates. Unless noted otherwise, analyses were run in SPSS 29 (Armonk, NY: IBM Corp). Significance was set at p ≤ .05.

Prevalence of recurrent HCU

Over a median follow-up of 10 years, half of the patients with NOCAD remained event-free (n=273), while 24% (n=130) had a single HCU and 26% (n=143) had multiple HCU. Among the latter, 59% had two events, 22% three, 13% four to six, and 6% seven to sixteen HCU. Descriptive information on the events per HCU group is visualized in Figure 1, and reported in supplemental Table S1, stratified by no HCU, single, and multiple HCU. Overall, 30% had ≥1 ED visit and 36% ≥1 retest. Mean time between events was 5.0 years (SD=1.0) for single HCU and 2.8 years (SD=1.0) for multiple HCU. MACE occurred in 2% of patients without HCU, 12% with single HCU, and 32% with multiple HCU. Cardiovascular mortality occurred in five patients (2%) without prior HCU. Mortality rates (cardiac and non-cardiac) did not differ significantly between HCU groups.

• Download figure
• Open in new tab

Figure 1. Number of events stratified for ED visit, Retest, and MACE per subgroup of health care utilization.

Descriptives stratified by HCU groups and covariate selection

No significant sex or age differences were found across HCU groups (Table 1). Patients with multiple HCU had higher anxiety, fatigue, poorer physical health, and lower quality of life. Moreover, the multiple HCU group had less college education, were more often obese, reported more chest pain, and were more often classified as INOCA.

Regularized regression identified education, obesity, and diabetes as consistent predictors of HCU, while other disease severity indicators and lifestyle/risk factors were not. These three predictors were added as covariates to the negative binomial GLM, along with age, sex, and MACE.

Number of HCU by mental health and health status

In separate models examining number of HCU, adjusted for age and sex, anxiety, fatigue, physical component, physical limitation, angina frequency, and quality of life were significantly associated with HCU count (Table 2). More anxiety, fatigue, and poorer health status at baseline were associated with increased HCU. No significant association with HCU was observed for depressive symptoms, Type D personality, mental component, angina stability, and treatment satisfaction. Subsequently, each model was adjusted for education, obesity, and diabetes, showing that angina frequency (B= −0.003, 95%CI −0.006 - −0.0001, p=.041), and quality of life (B= −0.004, 95%CI −0.006 - −0.001, p=.007) remained negatively associated with the number of HCU. A greater angina frequency and a lower quality of life at baseline were related to more HCU, as well as covariates having obesity or diabetes. Finally, in the models adjusted for MACE, MACE (B= 0.558, 95%CI 0.403-0.714, p<.001), quality of life (B= −0.003, 95%CI −0.006 - −0.0003, p=.031), and diabetes (B= 0.141, 95%CI 0.005-0.227, p=.042) were significantly associated with the number of HCU, but none of the other mental health or health status variables nor age, sex, college education, or obesity.

Explorative analyses: sex and MACE stratified findings

Explorative sex- and MACE-stratified associations between mental health or health status and HCU are presented in Supplemental Table S2. Among women, angina frequency and quality of life (modified SAQ) were associated with HCU, whereas no significant associations were observed in men. These sex-stratified models were adjusted for age, education, obesity, diabetes, and MACE. In women, both MACE and obesity, and in men, MACE, lower education, and diabetes were consistently associated with HCU (data not shown).

In patients without MACE, higher anxiety, poorer physical health, greater angina frequency, and lower quality of life at baseline were associated with higher HCU. In patients with MACE, higher depressive symptoms, more fatigue, and poorer physical and mental health status were associated with more HCU. These MACE-stratified models were adjusted for sex, age, education, obesity, and diabetes. Among covariates, obesity and diabetes were consistently related to higher HCU, but only in patients with MACE (data not shown).

Prevalence of HCU events in NOCAD

HCU was assessed as a composite outcome of all cardiac emergency department (ED) visits and repeat diagnostic cardiac tests over 10 years, to capture the recurrent and persistent nature of care use in this population. The high prevalence observed aligns with previous reports of persistent chest pain in NOCAD (5, 36).

In a retrospective cohort, Roguin and colleagues reported a lower recurrence (10% repeat CAG; 3% ischemic heart disease progression) with higher rates among men and those with cardiometabolic risk factors (37). In contrast, the present TWIST study prospectively included patients with any visible wall irregularities, representing a group with more pronounced coronary abnormalities. This broader inclusion yielded a higher prevalence (36%) of recurrent testing, not limited to CAG. Our results partly align with earlier studies showing repeat testing was more common among patients with minimal disease, diabetes, older age, and male sex (17, 38, 39). Overall, recurrent cardiac HCU is common in NOCAD and appears driven by a combination of clinical risk factors and psychosocial mechanisms explored below.

Mental health and health status at baseline predicting HCU

In age- and sex-adjusted analyses, higher anxiety and fatigue, poorer physical health, greater disease-specific limitation, more frequent angina, and lower quality of life were associated with greater HCU. After full adjustment, only lower quality of life remained an independent predictor. These findings align with prior research linking mental health and physical symptom burden to healthcare use in IHD (21, 23))(22). Experiencing depressive symptoms, in patients with obstructive CAD or who received cardiac surgery, was associated with increased ED visits and overall HCU (21, 23). A one-year follow-up of patients with IHD previously showed that anxiety, depression, and poorer cardiac-specific and generic health status were associated with higher healthcare costs (22). This emphasizes the intertwined role of psychological and physical functioning in care-seeking behaviour.

Explorative MACE-stratified analyses suggested that psychosocial correlates of HCU differ by disease progression. While anxiety, angina frequency, and lower quality of life were associated with HCU in patients without MACE, depressive symptoms, fatigue, and poorer physical and mental health were relevant in patients with MACE. Thus suggesting that psychosocial drivers of recurrent care may differ between patients with and without objective progression.

Most studies have examined QoL as an outcome variable. Our findings extend this literature by identifying it as a predictor of recurrent care use, independent of disease severity. Poor QoL may therefore represent a key driver of healthcare-seeking in NOCAD, beyond traditional risk factors or mental health symptoms.

Although no overall sex differences were observed, exploratory analyses suggested sex-specific patterns. In women, higher angina frequency and lower QoL predicted greater HCU, while no significant associations were found in men. This may reflect heightened symptom perception or concern about recurrence in the absence of clear disease markers, consistent with evidence linking anxiety to increased care-seeking in women with NOCAD and noncardiac chest pain (40, 41). Recent findings by Liu et al. (2024) support this interpretation. Among women with angina and no obstructive CAD (ANOCA), psychological distress was significantly elevated compared with controls, independent of objective ischemia (10). Notably, women with milder angina reported the highest levels of anxiety, depression, and perceived stress, suggesting that subjective symptom experience—rather than ischemic burden—may drive distress and recurrent care-seeking (10). Taken together, these findings point toward sex-specific psychosocial mechanisms underlying HCU in NOCAD, warranting further study and targeted psychosocial support.

Disease progression and covariates associated with HCU

Experiencing MACE was associated with greater HCU, as such events are often preceded by ED visits or diagnostic testing. However, a small proportion of patients (2%, n=6) experienced MACE without preceding HCU, including five who died from cardiovascular causes. Conversely, the majority of patients with one or multiple HCU episodes (88% and 68%, respectively) did not experience disease progression, highlighting that NOCAD burden extends beyond the pathway toward obstructive IHD. Recurrent HCU thus represents a meaningful dimension of patient burden.

Consistent with prior studies, baseline diabetes independently predicted higher HCU (37, 38). Obesity also showed an initial association, suggesting cardiometabolic factors contribute to symptom persistence and repeated testing. Vatsa et al. (2024) reported that obesity was associated with more frequent and severe chest pain, partly mediated by depressive symptoms (36), supporting this interpretation. Neither sex nor age were significantly related to HCU in our study, aligning with some (17, 37), but not all previous studies (37, 38).

Limitations

Several limitations should be considered. HCU was defined as cardiac ED visits or repeat diagnostic testing; other encounters, such as annual check-ups or arrhythmia assessments, were not included, potentially underestimating total healthcare utilization. Unlike studies defining hospitalization as admission ≥24 hours, our definition captured acute or diagnostic events possibly related to IHD progression. Each hospital contact was categorized as an ED visit or retest, irrespective of disease progression, which was analyzed separately.

Second, patient inclusion (2008–2013) preceded widespread use of coronary function testing for vasospasm or microvascular dysfunction (42, 43). Though ANOCA/INOCA could not be confirmed based on coronary function test, patients were categorized by presenting features (angina-like symptoms or detected ischemia on functional testing). The INOCA group was significantly associated with HCU compared to the ‘other’ group, but other covariates had a stronger association with HCU.

Finally, residual confounding and sample size constraints may have limited power in sex- and MACE-stratified analyses, despite the study’s 10-year longitudinal design.

Implications

Patients with signs of IHD but NOCAD remain a diagnostic and therapeutic challenge, contributing substantially to healthcare use. Poor baseline QoL independently predicted recurrent cardiac HCU, emphasizing the importance of addressing symptom burden and psychosocial factors in long-term management.

Recent evidence indicates that invasive coronary function testing can identify microvascular dysfunction or vasospasm, providing both prognostic and cost-effective benefits (44–46). The 2024 European Society of Cardiology Chronic Coronary Syndromes guidelines recommend such testing in patients with refractory angina and suspected ANOCA/INOCA to guide individualized treatment (47). Improving recognition and management of these conditions in both primary and specialist care—through better symptom assessment and early initiation of targeted therapy—may reduce recurrent healthcare use and improve patient well-being, even without invasive testing (8, 9).