1 Introduction

The Japanese Circulation Society/the Japanese Association of Cardiac Rehabilitation Joint Working Group (JCS/JACR) 2021 Guideline on Rehabilitation in Patients With Cardiovascular Disease have emphasized the importance of individual health literacy (HL) [1]. HL is defined as “the cognitive and social skills that determine the motivation and ability of individuals to gain access to, understand, and use information in ways which promote and maintain health.” [2] Patients with low HL have difficulties with complex health tasks and abilities, such as understanding health information and having limited access to healthcare; as well as expressing their concerns, emotions, and needs [3]. HL is considered to be essential when interacting with patients; however, medical staff overestimate the HL of patients in clinical practice, and only 14% of patients with low HL understand the content of patient education [4]. Low HL is correlated with poor lifestyle habits that increase the risk of cardiovascular diseases (CVD) during cardiac rehabilitation [3]. Furthermore, low HL increases mortality and hospitalization rates, while decreasing physical function and activities of daily living [3, 5, 6]. Increasing the level of knowledge of HL in patients results in improved HL-interventions and resultant health outcomes.

The concept of health-related quality of life (HRQOL) is correlated with the health and quality of life (QOL); in addition to the physical, psychological, and social well-being of the patient [7]. QOL is an all-inclusive concept encompassing all factors affecting a patient’s life. By contrast, HRQOL includes only those factors related to the patient’s health [8]. Cardiac rehabilitation and increased physical activity have been shown to improve HRQOL [9,10,11]. Moreover, the severity of disease, such as with an increasing New York Heart Association (NYHA) functional class, is associated with decreasing HRQOL. Therefore, disease preventive management is essential in controlling disease severity [11]. Collecting enough information about the disease is important in the enhancement of disease management and consequent HRQOL [12] HL involves the ability and skill to collect health information.

In a previous systematic review and meta-analysis, the HL status and HRQOL were found to be positively correlated in patients with CVD [13] These studies used disease-specific HRQOL assessments and the Short Form-12 Health Survey; however, the calculation of utility scores were not included in these tools. The utility scores of HRQOL are preference values attached to the overall health status of a patient [14] Utility scores are based on a scale ranging from 0 (death) to 1 (full health) [15, 16] and are useful in the decision-making process regarding choosing an appropriate healthcare intervention or treatment [14] However, the correlation between HL and utility-based health-related quality of life (HRQOL) scores has been poorly documented. In the present study, we aimed to examine the correlation between HL and utility-based HRQOL scores in study participants undergoing cardiac rehabilitation. Moreover, based on a comprehensive questionnaire, we hypothesized that study participants with low HL would have lower HRQOL.

2 Methods

2.1 Study design and eligibility criteria

This study was a cross-sectional multicenter clinical trial named, “The Kobe-Cardiac Rehabilitation Project for People Around the World (K-CREW)”. The K-CREW involved four small- and medium-scale hospitals with 200–580 beds. The number of beds was allotted based on the independent criteria of each hospital for the implementation of cardiac rehabilitation in Japan. We included patients who were admitted to these affiliated hospitals between October 1, 2020, to March 31, 2023. The inclusion criteria were as follows: patients with various principal diagnoses who had participated in cardiac rehabilitation; had no short-term hospitalization for medical consultation, such as admission for coronary angiography (CAG), percutaneous coronary intervention (PCI), ablation, and pacemaker battery replacement; and had been hospitalized for more than 5 days. Additionally, the exclusion criteria were as follows: patients with diagnosed, or probable dementia as determined by a Mini-Mental State Examination (MMSE) score < 24; inability to walk independently; refusal of informed consent for participation in the study, hospital mortalities; and clinical records comprising data deficits.

First, 9880 patients with CVD were hospitalized during the study period. After adjusting for the inclusion criteria, 3068 patients who participated in cardiac rehabilitation were included. Finally, 448 patients, subsequently referred to as participants after signing informed consent, were included in the statistical analysis, after the application of the exclusion criteria to the remaining patients (Fig. 1).

Fig. 1
figure 1

Flow diagram depicting the selection of study participants. This figure depicts the patient flow from the initial population of patients to the final analysis of participants after informed consent was obtained. Many patients were admitted for short periods without rehabilitation and were not included in the final analysis

Full size image

The participants included, in the final analysis, were divided into the following four sub-cohorts: the “heart failure (HF),” “ischemic heart disease (IHD),” “valve disease,” and “other” cohorts, “Other” referred to conditions such as aortic dissection, aortic aneurysm, and infective endocarditis.

2.2 Phase I cardiac rehabilitation

Phase I cardiac rehabilitation is an in-hospital program consisting of exercises and education [17] Cardiac rehabilitation is initiated within 3 days after admission for cardiac surgery. According to the JCS/JACR 2021 Guideline on Rehabilitation in Patients With Cardiovascular Disease [1], the exercise program includes aerobic exercise and resistance training for 20–40 min/day, 5–7 days/week. In this study, the aerobic exercise program differed slightly among the affiliated hospitals and was implemented for up to 25 min/day. We adjusted the exercise intensity according to either the Borg rating of perceived exertion (RPE) scale ranging from an RPE of 11 to 13, or the anaerobic metabolism threshold of the participant [18] Resistance training targeted limb muscles for mainly 10–20 min. Participants education included lectures on disease management and lifestyle modification by a multidisciplinary team of doctors, nurses, registered dietitians, pharmacists, physical and occupational therapists, and health and fitness instructors.

2.3 Assessment of health literacy

The 14-item Health Literacy Scale (HLS-14) questionnaire [19] was used to evaluate HL. The HLS-14 questionnaire had 14 items, using a 5-point Likert scale, where 1 point was indicative of the participant strongly agreeing, while 5 points indicated that the participant strongly disagrees. This questionnaire was administered to participants in Japanese. HL was divided into three subclasses: functional, communicative, and critical. Functional HL refers to basic skills in reading and writing, while communicative HL refers to the skills required to participate in everyday activities, extract information, and apply new information. Critical HL refers to more advanced skills used in the critical analysis of information. The HLS-14 questionnaire has demonstrated adequate reliability and validity (Cronbach’s α: 0.81) [19].

2.4 Assessment of HRQOL

We used the EuroQol 5-Dimension 5-Level (EQ-5D-5L) questionnaire [20] to assess the HRQOL of the participants, which is a generic instrument for describing and evaluating health. The EQ-5D-5L questionnaire is one of the most commonly used measurement systems. [21] The EQ-5D-5L questionnaire has five dimensions: mobility, self-care, usual activities, pain or discomfort, and anxiety or depression. In addition, this questionnaire has five severity levels: no, slight, moderate, severe, and extreme problems/ unable to. The participants answered each of the five dimensions according to the five severity levels on the questionnaire. The results of the EQ-5D-5L questionnaire were converted into utility scores. We used the Japanese EQ-5D-5L value set to calculate the utility scores [22] that ranged from -0.025 to 1.000 (full health).

2.5 Baseline characteristics and other variables

We extracted the following baseline characteristics from the medical records, at the time of admission: age, sex, body mass index (BMI), employment, whether the participant lived with someone else, comorbidities, and smoking and marital status. Additional variables included the estimated glomerular filtration rate (eGFR); Geriatric Nutritional Risk Index (GNRI) [23]; Charlson Comorbidity Index (CCI) [24]; as well as the serum hemoglobin, albumin, white blood cell count, and brain natriuretic peptide (BNP) levels. Nutritional risk categories, as determined by the GNRI, were as follows: major (GNRI < 82), moderate (82 ≤ GNRI < 92), low (92 ≤ GNRI ≤ 98), and no risk (GNRI > 98) [23] At discharge, up to three cardiac rehabilitation healthcare professionals evaluated the HLS-14, EQ-5D-5L questionnaires, and cognitive function. Furthermore, one researcher obtained the prescription data from the medical records.

2.6 Statistical analysis

We conducted a comparative analysis of the low and high HL cohorts. First, participants were divided into low and high HL cohorts, with a cutoff of 50 points in HLS-14, as per the original HLS-14 study [19] The normality of each variable was checked using the Shapiro–Wilk test. Based on the results, the Student’s t-test or Wilcoxon rank-sum test was used for continuous variables. Pearson’s chi-squared or Fisher’s exact tests were used for categorical variables. Figure 2 summarizes the response distribution of the EQ-5D-5L questionnaire for the low and high HL cohorts.

Fig. 2
figure 2

Distribution of EQ-5D-5L questionnaire scores in the low and high health literacy cohorts are shown. This figure shows the percentage distribution of responses for each severity level for each dimension of the EQ-5D-5L questionnaire. For approximately all dimensions of the EQ-5D-5L questionnaire, both the low and high HL cohorts responded positively to the severity level of “no problem.” However, the high HL cohort was more likely to respond positively to the severity levels of “no problem” and “extreme problems/ unable to” than the low HL cohort. EQ-5D-5L EuroQol 5-Dimension 5-Level, HL health literacy

Full size image

Generalized linear mixed models for the utility scores of the EQ-5D-5L questionnaire were calculated, and sub-cohort analyses were conducted. Because the present study used cluster sampling, the institutional identifier of the affiliated hospitals was set as the random effect. Univariate and multivariate analyses were performed. The following independent variables were selected based on previous studies: age, sex, BMI, education, employment, the participant living with someone else, smoking and marital status, duration of admission, heart failure (HF), ischemic heart disease (IHD), valve disease, congestive cardiac failure (CCF), diabetes mellitus, cerebrovascular diseases (CVD), renal disease, mild cognitive impairment (MCI); as well as the GNRI and HLS-14 scores [25,26,27] HF, IHD, and valve disease were the principal diagnoses leading to admission. Sub-cohorts were created according to sex and the principal diagnoses based on previous studies [3, 28, 29] First, models were created for the male and female sub-cohorts, with the exclusion of the variable, sex. Second, the HF and IHD sub-cohort models were calculated, with the exclusion of valve disease.

Analyses were conducted without missing data for all variables. Nonparametric, continuous data were expressed as median values and interquartile ranges. Parametric, continuous data were expressed as means ± standard deviations. We set the significance level to p < 0.05 and used R version 4.3.0 [30]

2.7 Ethics approval and informed consent

The ethics committee of Kobe University (no. 951–1) approved the K-CREW project on August 12, 2020. Furthermore, approval for the conduct of this study was obtained from the ethics committees of each affiliated hospital (Sakakibara Heart Institute of Okayama, Sanda City Hospital, Shinyukuhashi Hospital, and Yodogawa Christian Hospital). This study was conducted in accordance with the principles of the Declaration of Helsinki [31] Written and signed informed consent was obtained from each participant deemed eligible.

3 Results

3.1 The overall trends

The median age was 71.0 [61.0–78.0] years old, and 75.7% of participants were male. The duration of hospitalization was 15 [12.0–20.0] days, and 98.9% of participants were discharged home. The principal diagnoses, HF, IHD, and valve disease, were 30.6%, 56.9%, and 5.8%, respectively. Other diseases included aortic dissection, aortic aneurysms, and infective endocarditis. The median utility measure was 0.88 [0.75–1.00]. More than 60% of participants positively responded to the severity level of “no problem” in each dimension of the EQ-5D-5L questionnaire.

3.2 Comparative analysis of the low and high HL cohorts

Of all the participants, 60.0% (n = 269) had low HL. The mean HLS-14 scores were 42.2 ± 5.6 and 55.7 ± 4.1 in the low-HL and high-HL cohorts, respectively. A comparative analysis of the low and high HL cohorts revealed statistically significant differences in the following variables: age, MCI, angiotensin-converting enzyme (ACE) inhibitor use, and utility scores (Table 1). The low HL cohort was older and had a greater frequency of MCI than that of the high HL. Furthermore, the high HL cohort was prescribed ACE inhibitors more often and had higher utility scores than that of the low HL cohort. The utility scores were 0.87 [0.74–1.00] and 0.89 [0.78–1.00] for the low and high HL cohorts, respectively (p-value = 0.03). Figure 2 shows the percentage distribution of responses for each level and dimension of the EQ-5D-5L questionnaire (Fig. 2). For approximately all dimensions of the EQ-5D-5L questionnaire, both the low and high HL cohorts responded positively to the severity level of “no problem.” However, the high HL cohort was more likely to respond positively to the severity levels of “no problem” and “extreme problems/ unable to” than the low HL cohort.

Table 1 Baseline characteristics of the low and high health literacy cohorts
Full size table

3.3 Multivariable analysis of the utility scores

In the generalized linear mixed model, the statistically significant explanatory variables for the utility scores were sex, education, marital status, and duration of admission (Table 2). This analysis revealed that utility scores were higher in male participants who were married, had a greater academic degree than a college education, and had a shorter duration of admission. HLS-14 scores were not found to be statistically significant explanatory variables for the utility scores, with a regression coefficient of 0.001 (95% CI, -0.003 to 0.005). The adjusted R2 value of the model was 0.36.

Table 2 Generalized linear mixed model for utility scores of the EuroQol 5-Dimension 5-Level (EQ-5D-5L) questionnaire
Full size table

Moreover, sub-cohort analyses were conducted to determine the proportion of the participants that were male or female, and whether the foremost principal condition was HF or IHD. However, the HLS-14 scores were not statistically significantly correlated with the utility scores, in any of the sub-cohort analyses (Table 3).

Table 3 Sub-cohort analysis of the generalized linear mixed model for utility scores of the EuroQol 5-Dimension 5-Level (EQ-5D-5L) questionnaire by sex and principal diagnosis
Full size table

4 Discussion

4.1 Summary

This novel study is the first to have found a correlation between HL and EQ-5D-5L questionnaire scores in participants undergoing cardiac rehabilitation. The overall median HRQOL was 0.88, with most participants responding positively to the severity level of "no problem" on the EQ-5D-5L questionnaire. In addition, the utility scores in the high HL cohort were statistically significantly higher than those of the low HL cohort. However, the results of the sub-cohort analyses of the generalized linear mixed model by sex and principal diagnosis revealed that the HLS-14 scores were not statistically significantly correlated with the utility scores.

4.2 Comparison with previous studies

A systematic review and meta-analysis included three cross-sectional studies; two included patients with HF and one included patients with IHD [13] The included studies were published between 2011 and 2020; and were conducted in the USA, Australia, and China. The mean age was 60.7–72.0 years old; and in three studies, the rates of low HL ranged from 14.3 to 52.5%. By contrast, the present study was a cross-sectional study that included participants with HF, IHD, and valve disease. The median age was 71.0-years-old, and the rate of low HL was 60.0%. Thus, this study included patients with multiple CVD comorbidities who had targeted cardiac rehabilitation and had a higher rate of low HL than that of previous studies. Regarding regional differences, Japanese populations tend to exhibit lower health literacy compared to other regions, such as Europe. [32] Despite variations in health literacy assessment methods, our present study revealed a higher percentage (60.0%) of patients with low health literacy than the previous review. [13]

The median utility score of the present study was 0.88, and the "no problem" response rate of the EQ-5D-5L questionnaire was high. By contrast, in another systematic review and meta-analysis, the mean utility score was 0.77 [29] EQ-5D-5L questionnaires with high utility scores of 0.87–1.00 are correlated with lower mortality risks than EQ-5D-5L questionnaires with low or middle utility scores [33] Therefore, the participants of the present study who had utility scores closer to 1.00 (full health), may have had a better prognosis. This was evidenced by the rate of participants who were discharged home which was approximately 100%.

This study showed that the utility scores of patients with CCF in Asia were better than those of patients in other regions [12] Zhou et al. (2021) reported that male patients had better HRQOL as measured by the EQ-5D-5L questionnaire than female patients. The severity of the disease is reflected by the magnitude of the utility score. As the disease worsens, the utility score decreases. Furthermore, the utility score is influenced by the personal characteristics and living environment of the patient. Similarly, the multivariate analysis of the present study revealed that utility scores were higher in male participants.

Consequently, it is inadequate to consider regional differences alone. We should focus on the personal characteristics, as well as the family and cultural environments of patients to fully comprehend their HRQOL [21]

4.3 Clinical implications

HRQOL assessments involve disease-specific and comprehensive methods. Many disease-specific assessments were used in a systematic review and meta-analysis involving patients with CCF [12] Disease-specific assessments include the Minnesota Living with Heart Failure Questionnaire (MLHFQ) [34] and the Improving Chronic Illness Care Evaluation (ICICE) Heart Failure Symptom Scale (HFSS) [28] Disease-specific assessments have often been used in studies on patients with HF and have focused on physical and mental disorders associated with psychosomatic symptoms. In addition, the MLHFQ has been used to assess the treatment response of HF interventions [35] However, these assessments are limited in the evaluation of other diseases. Moreover, the external validity is difficult to determine.

The EQ-5D-5L questionnaire is a comprehensive HRQOL assessment. These assessments are more practical, due to their higher completion rates and lower response burdens [33] The EQ-5D-5L questionnaire can be used to calculate the utility scores of several diseases. However, the limitation of comprehensive questionnaires is that they are not very discriminative or responsive [36] The EQ-5D-5L questionnaire has been reported to be affected by the ceiling effect [37] The novelty of the present study was to verify the correlation between HL and utility-based HRQOL scores using the comprehensive EQ-5D-5L questionnaire. However, the findings of this study did not show positive correlations as with the results of previous studies. Considering the advantages and disadvantages of comprehensive and disease-specific assessments, we believe that disease-specific assessments may be more sensitive to HRQOL in patients with CVD.

The systematic review and meta-analysis by Kanejima, et al. included studies showing a positive correlation between HL and HRQOL. In the current study, a comparative analysis of the low and high HL cohorts showed a statistically significant difference in the utility scores of the EQ-5D-5L questionnaire. However, multivariate analyses revealed no statistically significant positive correlation between HL and HRQOL, which may have not extended to all patients undergoing cardiac rehabilitation. Based on these results, confounding and intermediate variables may have influenced HL and HRQOL. Sex and the principal diagnosis substantially affect HRQOL. [3, 28, 29] Sex and the principal diagnosis were adjusted for sub-cohort analyses in this study; nonetheless, they were not valid confounding or intermediate variables affecting the correlation between HRQOL and HL. Future studies should investigate other confounding or mediating variables influencing the correlation HRQOL and HL, such as self-efficacy and self-care [38] Furthermore, there is a positive correlation between self-care and HL [39] In the current study, 75.8–79.3% of participants responded positively to the severity level of “no problem” for self-care in the EQ-5D-5L questionnaire, while 60% had low HL. According to a previous study, patients with low HL experience problems with self-care. Patients with low HL may lack knowledge about their disease. Further investigations of the correlations between HL and HRQOL would assist in the prevention of incorrect interventions or identify patients for whom HL interventions are more effective.

4.4 Study limitations

The present study had several limitations. First, the patients had multiple CVD comorbidities. While there was value in providing an overview of cardiac rehabilitation, we may not have captured the characteristic of each comorbidity. Second, only the data of participants with MCI were analyzed in the study; therefore, the results of the questionnaires might have low reliability. Third, the HLS-14 and EQ-5D-5L questionnaire used were assessments with less objectivity and may have had ceiling effects. Fourth, this study did not include variables associated with HL, such as economic status. Finally, this study may have had selection bias. The eligibility criteria excluded many patients with short-term hospitalizations. Moreover, the EQ-5D-5L questionnaire included mobility as one of the dimensions. The exclusion criterion in this study was the inability to walk independently; therefore, only participants who could walk independently were included. Thus, the number of mobile participants who responded positively to the severity level of "no problem" was higher. The criteria, not found in previous studies may have been strict, leading to higher utility scores owing to selection bias.

5 Conclusion

This study investigated the correlation between HL and utility-based HRQOL scores in participants undergoing phase I cardiac rehabilitation. We hypothesized that the utility-based HRQOL scores would be substantially lower in the low HL cohort than in the high HL cohort. Overall, the median HRQOL score was 0.88, and many participants responded positively to the severity level of "no problem" on the EQ-5D-5L questionnaire. The results of the comparative analysis of the low and high HL cohorts supported the research hypothesis. However, multivariate analysis revealed that the HLS-14 scores were not statistically significantly correlated with the utility scores. Confounding or intermediate variables may lead to increased HL and utility-based HRQOL scores; however, sub-cohort analyses according to sex and the principal diagnosis did not reveal any statistically significant results. Although this study used the EQ-5D-5L as a comprehensive questionnaire, disease-specific questionnaires may better capture the correlation between HL and HRQOL. Future studies should be conducted that further explore the confounding or intermediate variables of HL and HRQOL.