Hypothetical model of perceived adherence to treatment among patients with coronary heart disease after a percutaneous coronary intervention

Abstract Aim To test the hypothetical model of adherence to treatment among patients with coronary disease after percutaneous coronary intervention. Design A descriptive, explanatory, cross‐sectional survey. Methods The study was conducted in 2013 with 416 patients in five hospitals in Finland. The adherence of patients with chronic disease instrument, the adherence visual analogue scale, the social support for people with coronary heart disease instrument, the EuroQoL five‐dimensional scale and EuroQoL visual analogue scale were used. The data were analysed using descriptive statistic. The hypothetical model was tested using structural equation modelling. Results The hypothetical model explained 30% of perceived adherence to treatment. Structural equation modelling confirmed that motivation, support from physicians and next of kin had direct associations with adherence. Indirectly, informational support, results of care, perceived health, anxiety and depression were associated with adherence. The background variables associated with adherence were gender, relationship, physical activity, consumption of vegetables and consumption of alcohol.

The use of PCI has increased steadily over the past decade.
When the treatment is successful, a patient may even be discharged on the same day as the procedure. Shorter hospitalizations are clearly cost-effective (Rubimbura et al., 2019;Shroff et al., 2016), but the responsibility for care transfers quickly to patients, which may lead to a diminished understanding of the seriousness of CHD and the lifelong chronic nature of the disease (Ibanez et al., 2017).

| Background
The Theory of Adherence among People with Chronic Disease (1999) is originally developed by Kyngäs (1999) and developed further among different patient groups (Kähkönen et al., 2015;Oikarinen, Engblom, Kääriäinen, & Kyngäs, 2017). According to this theory, adherence to treatment is defined as an active, intentional and responsible process of care where patients with chronic conditions work to maintain their health in collaboration with healthcare professionals. Adherence to treatment consists of adherence to medication and a healthy lifestyle, including a healthy diet, physical activity, non-smoking and moderate alcohol consumption.
Explanatory factors associated with adherence to treatment according to Kyngäs's original Theory of Adherence among People with Chronic Disease are responsibility, cooperation, a sense of normality, motivation, the results of care, support from next of kin, support from nurses, support from physicians and a fear of complications (Kähkönen et al., 2018;Kyngäs, 1999;Oikarinen et al., 2017). The Theory of Adherence among People with Chronic Disease as a theoretical framework to evaluate post-PCI patients' adherence has been tested. Cooperation, a sense of normality, motivation, the results of care, support from next of kin, support from physicians and a fear of complications have been shown to explain adherence to treatment among patients with CHD after PCI (Kähkönen et al., 2018). Post-PCI patients' non-adherence to treatment may be intentional or unintentional. Intentional non-adherence is associated with the patient's conscious decision to stop taking the prescribed medicines, reduce the dosage of one's medication or neglect treatment recommendations. The reasons for intentional non-adherence can be numerous such as financial obstacles or medication side effects. In contrast, unintentional non-adherence refers to a patient's lack of capacity or cognitive resources, which can lead to non-adherence. (Molloy et al., 2014;Pettersen et al., 2018).
Certain psychological factors, such as poor perceived health (De Smedt et al., 2016;Sajobi et al., 2018) and a lack of social support (Valtorta, Kanaan, Gilbody, Ronzi, & Hanratty, 2016;Xia & Li, 2018), are evidently associated with the prognosis of CHD (De Smedt et al., 2016;Sajobi et al., 2018;Valtorta et al., 2016). Perceived health reflects an individual's overall and subjective perceptions of his or her own health status, which include physical and psychological aspects (De Smedt et al., 2014). Perceived health predicts mortality and morbidity (Sajobi et al., 2018) despite the manifestation of other co-morbidities and risk factors among patients with CHD. Hence, perceived health is an important outcome measure among patients with CHD. The significance of patients' subjective experience of their own health status in clinical practice in undeniable (De Smedt et al., 2016;Sajobi et al., 2018).
Furthermore, a low level of social support is considered a risk factor for CHD in healthy people, as well as a risk factor for worse prognoses and higher mortality among patients diagnosed with CHD (Valtorta et al., 2016;Xia & Li, 2018). The dimensions of social support, such as support from next of kin, physicians and nurses, have been proven to be significant factor associated with adherence among chronically ill patients (Kähkönen et al., 2018;Kyngäs, 1999). Social support has been defined as a dynamic interpersonal process centred on the reciprocal exchange of information. Social support changes across contexts, and it is manifested between providers and recipients. Social support may appear multifaceted depending on its context. (Brooke & Collins, 2015).
In particular, the importance of social support is clear during a stressful life event, such as an acute cardiac event. Social support provides security, a feeling of love and a community spirit (Aazami, Jaafarpour, & Mozafari, 2016).
These concepts are evidently associated the prognoses of patients with CHD (De Smedt et al., 2016;Valtorta et al., 2016). Although adherence to treatment, perceived health and social support have been widely studied, an absence of previous research on the relationships between these theoretical concepts is undeniable. In fact, CHD is a major public health problem worldwide, causing a significant financial and human burden. Therefore, improved knowledge is needed to increase our theoretical understanding of adherence to treatment and associated factors to develop effective nursing interventions that meet current treatment needs.

| Aim
The aim of the present study was to test whether empirical data would fit the proposed hypothetical model of perceived adherence to treatment (Figure 1) based on four sub-studies: testing the theory of adherence to treatment of chronically ill patient among patients (Kähkönen et al., 2015), description and exploration of the predictors of adherence to treatment (Kähkönen et al., 2018), perceived health (Kähkönen, Saaranen, Lamidi, Miettinen, & Kankkunen, 2017, received social support (Kähkönen, Kankkunen, Miettinen, Lamidi, & Saaranen, 2016) and associated factors among patients with CHD after PCI. The hypothetical model of perceived adherence to treatment was created based on the statistically significant findings in these sub-studies. The major hypothesis of this study was that the hypothetical model of perceived adherence to treatment is suitable for explaining adherence to treatment and associated factors among patients with CHD after PCI. To test the hypothetical model of perceived adherence to treatment, the following specific hypotheses were set: 1. Based on the tested Theory of Adherence of Patients with Chronic Disease among post-PCI patients: sense of normality, cooperation, motivation, support from next of kin, results of care, support from physicians and fear of complications predict perceived adherence to treatment.
2. Dimensions of social support: informational, support and functional support predict perceived adherence to treatment.
3. Perceived health and its dimension: problems in mobility, usual activities, pain, discomfort, anxiety and depression predict perceived adherence to treatment.
4. Sociodemographic (gender, age, relationship, length of education, employment status, profession), health behavioural (consumption of vegetables, alcohol consumption, physical activity, smoking) and disease-specific factors (LDL cholesterol, total cholesterol, systolic blood pressure, duration of CHD, pervious PCI, AMI or CABG) predict with perceived adherence to treatment.

| Design
This is the combination study which tests the hypothetical model based on the results of descriptive, explanatory, cross-sectional surveys regarding the level of adherence to treatment, received social support and perceived health of patients with CHD after PCI.

| Participants
A convenience sample of patients with CHD was recruited into the study 4 months after PCI, when the patients had recovered from treatment and adapted to everyday life. Patients were drawn from a multicentre arrangement including two university hospitals and three central hospitals in Finland. The inclusion criteria were an age of 18-75 years, PCI having been conducted in under elective or acute circumstances and fluency in the Finnish language. Patients diagnosed with a memory disorder were excluded from the study. The potential participants' suitability for participation was evaluated, and 572 patients met the inclusion criteria and received verbal and written information about the study from the registered nurse of the medical ward during the hospitalization. After that 91% (520) agreed to participate and provided informed consent. According to power analyses, this sample size was large enough to detect statistical significance with a power of 80% and a significance level of 0.05 given relatively small correlations (0.14). This number of observations and incidence rate can detect about 7%-13% of the difference between the groups.

| Data collection
The data were collected using a postal questionnaire 4 months after PCI between January and December 2013 via the following four instruments: 1. The adherence of patient with chronic disease (ACDI) instrument, which is based on a theoretical model of chronically ill F I G U R E 1 Hypothetical model of perceived adherence to treatment and related factors. Abbreviations: AMI, acute myocardial infarction; CABG, coronary artery bypass crafting; CHD, coronary heart disease; LDL, low-density lipoprotein; PCI, percutaneous coronary intervention

Expalnatory factors on adherence to treatment based on the Theory of adherence among Chronically ill PaƟents and predistors of adherence to treatment
• Sense of normality patients developed and tested originally by Kyngäs (1999). The ACDI contained 37 items of adherence to treatment, which were rated on a 5-point Likert scale ('definitely disagree' to 'definitely agree').

The adherence visual analogue scale (A-VAS) instrument, which
records the respondents' self-rated adherence to treatment from the best imaginable adherence to treatment (100) to the worst imaginable adherence to treatment (0)

The EuroQoL visual analogue scale (EQ-VAS) ranking respond-
ents' perceived health with endpoints labelled the best imaginable health (100) and the worst imaginable health state (0).
Additionally, the questionnaire included 18 self-reported background questions related to demographic details (age, gender, relationship, profession, employment status and length of education), disease-specific information (duration of CHD, previous AMI, previous PCI, previous CABG, systolic and diastolic blood pressure, total cholesterol and LDL cholesterol) and health behaviour (physical activity, consumption of vegetables, smoking habits and alcohol consumption). In summary, the questionnaire consisted of 56 items, which were rated on a 5-point Likert scale, two visual analogue scales (VAS) ranking 0-100 and 18 background questions.
Of the recruited patients with CHD after a PCI, 80% (n = 418) completed the study. Two questionnaires were rejected because they were inadequately completed and had missing values.

| Ethical considerations
The statement of the Ethical Review Board of the university hospital   in sub-studies I-IV were examined using structural equation modelling (SEM) because these specific phenomena and the relationships between the involved concepts have not previously been studied together. (Kellar & Kelvin, 2013). Statistically significant associations between the ACDI, A-VAS, EQ-VAS and SSCHDI instruments were found and these are presented in Figure 2. Therefore, it can be assumed that these instruments can be used to evaluate perceived adherence to The correlation between the hypothesized model of perceived adherence to treatment and the observed correlation matrix was examined using SEM. The chi-square test and its derivatives, CFI and RMSEA were used to estimate the adequacy of the model. A weak effect was indicated by values <0.10, a medium effect was indicated by values ~0.30, and a major effect was indicated by values >0.5.

| Validity, reliability and rigour of instruments
The construct validity of ACDI and SSCHD was verified with an exploratory factor analysis (EFA). In the ACDI, the alpha coefficients ranged from 0.40 to 0.88 and the alpha coefficient of the entire scale was 0.80, indicating sufficient-to-high internal consistency (Polit & Beck, 2016) as seen also in the earlier studies (Oikarinen et al., 2017;Ylimäki, Kanste, Heikkinen, Bloigu, & Kyngäs, 2015). Regarding ACDI, eleven factors explained 65% of the total variance, communalities varied between 0.20 and 0.80 and the factor loadings were between 0.30 and 0.90 (Polit & Beck, 2016).
A-VAS was developed for the present study. Therefore, the data of validity and reliability were not available. SEM is valuable data analysis tool in testing theories and in evaluating construct validity of the instruments (Kellar & Kelvin, 2013). Therefore, the validity was tested between the A-VAS and the adherence items in the ACDI (adherence to medication and adherence to a healthy lifestyle), which has been used widely measuring adherence among different patient groups. ACDI is based on the theory of adherence to treatment among chronically ill patients (Kyngäs, 1999). A statistically significant correlation using SEM was found between A-VAS and adherence items in the ACDI (0.50).
Additionally, EFA was conducted to verify the construct valid- The EQ-5D instrument is a widely used, standardized and validated generic instrument for measuring perceived health. In the earlier studies, the EQ-5D appears to have good convergent and known-groups validity (Yfantopoulos & Chantzaras, 2017).
Furthermore, the questionnaire's face validity was evaluated by three nurses who had extensive experience nursing cardiac patients and 15 patients with CHD after PCI. Based on their feedback, changes were made to increase the questionnaire's intelligibility and usability.

| RE SULTS
The study participants were mainly male CHD patients (75.5%, N = 314) aged from 18 to 75, with a mean age of 63.2 years (range In the eventual model (Figure 3), the strongest association was found between perceived adherence to treatment-associated and motivation. Additionally, perceived adherence to treatment was directly associated with support from physicians and next of kin. An indirect but statistically significant association was found between perceived adherence to treatment and informational support, the results of care and perceived health, as measured via motivation. Furthermore, anxiety and depression and perceived health had a statistically significant association. Regarding background variables, indirect associations were found as follows: the close personal relationship was associated with support from next of kin, support from physicians was associated with alcohol consumption and anxiety and depression, previous PCI and consumption of vegetables were associated with the perceived F I G U R E 3 Testing the hypothetical model including the factors directly and indirectly associated with adherence to treatment among patients with coronary heart disease after percutaneous coronary intervention. The effects of the standardized estimates were evaluated as follows: the estimates had a weak effect if the values were <0.10, medium effect if the values were ~0.30 and they had a major effect if the values were >0.5. Covariances: informational support-support from physicians***, physical activity-informational support*, consumption of vegetables-physical activity***, consumption of vegetables-gender*** (Note: *p < .05, **p ≤ .01, ***p ≤ .001). (Shreiber et al., 2006)  results of care, male gender was associated with better perceived health and physical activity was associated with motivation and anxiety and depression. Nevertheless, the outlined structural equation model with standardized estimates was statistically significant, with the following estimates: χ 2 = 908.1, df = 365, p < .001, χ 2 /df = 2.5, CFI = 0.8 and RMSEA = 0.6. Therefore, variables with ≥4 items were analysed using a unidimensional test. Considering the results, one item (16) was removed from the mean sum variable support from next of kin. In consequence, the outlined structural equation model indicated an acceptable model fit, with the following standardized estimates: CFI = 0.9 and RMSEA = 0.6. The chi-square test results were as follows: χ 2 = 786.7, df = 338, p < .001 and χ 2 /df = 2.4. The standardized path coefficients indicated a medium effect (~ 0.2-0.4) regarding the factors directly affecting adherence to treatment. (Shreiber et al., 2006;Bollen et al., 2014). The model explained 30% of the variance in the factors associated with adherence to treatment.

| D ISCUSS I ON
According to our knowledge, the theoretical model of perceived adherence has not been tested previously among post-PCI patients.
An absence of previous research on the relationships between the theoretical basis of perceived adherence to treatment is undeniable, although perceived health, social support and adherence to treatment have been widely studied. The present study indicated direct associations with perceived adherence to treatment and motivation, support from physicians and next of kin.
The strongest predictor to perceived adherence was motivation, whereupon it is essential to examine the factors underlying motivation to develop the nursing interventions needed to support patients' adherence to proper medical treatment and a healthy lifestyle, which are the cornerstones of a good prognosis of CHD after PCI (Ghisi 2015;Ibanez et al., 2017;Shroff et al., 2016). The results of the present study confirmed that perceived health and results of care predicted motivation. Therefore, patients should know and understand the special characteristic of CHD and the chronic long-term nature of CHD after PCI, which can help him or her recognize the importance of the results of care and perceived health. This requires informational support, which was also associated with motivation. These findings emphasize the importance of person-centred care, which necessitates collaboration between practitioners and patients, as well as the patients' expert participation in their own care. Additionally, physical activity was a predictor to motivation, as were anxiety and depression negatively. The benefits of physical activity are strongly documented in that it reduces modifiable CHD risk factors, such as hypercholesterolaemia, hypertension, overweight and stress, thus improving patients' prognoses (Perk et al., 2015). Therefore, it is reasonable to emphasize the importance of patient education regarding physical activity so that patients are informed about its value (Edward et al., 2016;Ghisi et al., 2015;Perk et al., 2015). Support from physicians was also directly associated with perceived adherence to treatment. At present, health care in Finland is undergoing a significant structural change, which is strongly associated with a reassessment of duties, including the transfer responsibility from physicians to nurses concerning chronically ill patients' Furthermore, one important finding was that support from the next of kin was associated with perceived adherence to treatment.
This result emphasizes the importance of the next of kin's participation in a patient's care and education. The goal of the next of kin should be to support and encouragement that is in accordance with the patient's wishes. Risk factors for CHD are generally associated with health behaviours, such as diet and physical activity. Adherence to a healthy lifestyle requires changes in health behaviours after PCI, which will be easier to make if the next kin understands the importance of adherence to a healthy lifestyle and supports the patient.
Being married or in a close relationship was associated with receiving support from the next of kin. (Lammintausta et al., 2014). This is noteworthy because marriage or being a close relationship, which enables patients' feelings of love, belonging to a community and security in life, has a protective effect on CHD patients' prognoses according to prior evidence (Xia & Li, 2018

| Limitations
This study includes certain theoretical and methodological limitations. Although perceived health, social support and adherence to treatment have been widely studied, an absence of previous research on the relationships between these main concepts complicated the theoretical basis for this study. Additionally, the fragmentation of these concepts was challenging in terms of comparing various studies. This fragmentation may also cause some significant aspects of the phenomena to go unidentified.
In terms of methodological limitations, the hypothetical model which was tested in this study may have some limitations regarding causality and generalizability. However, it is not the main concern in the present study, because the hypothetical model based on the three sub-studies, which were verified previously (Kähkönen et al., 2016(Kähkönen et al., , 2015(Kähkönen et al., , 2018(Kähkönen et al., , 2017. However, A-VAS was developed for this study and the reliability and validity of this instrument have not been tested previously, which could be a limitation in this study. Regarding the questionnaire, the number of the total question was quite high, which could have been a burden to some patients and influencing the accuracy of the answers provided by the patients. Although the response rate was good (80%), this may be a limitation.
Finally, in self-reported data collection methods, there is always a risk of the social desirability effect, where patients provide answers, they think are favourable instead of saying what they do or think (Althubaiti, 2016).

| CON CLUS ION
The tested hypothetical model was confirmed in part in this study.
First, the Theory of Adherence among People with Chronic Disease is an applied suitable theoretical framework for evaluating adherence to treatment among patients with CHD after PCI: motivation, support from physicians and support from next of kin were the strongest predictors of perceived adherence to treatment and results of care predicted adherence indirectly. Instead, sense of normality, cooperation and fear of complication were not associated with perceived adherence to treatment. Second, perceived health and its dimension anxiety/depression were indirect predictors to perceived adherence to treatment. Other dimensions of perceived health: mobility, usual activities and pain/discomfort were not associated with perceived adherence to treatment. Third, informational support as a dimension of social support predicted perceived adherence to treatment via motivation, but emotional support and functional support were not associated with perceived adherence to treatment. Regarding background factors, physical exercise was predicted motivation and was associated with lower anxiety/depression. The close relationship was associated with support from next of kin. Instead, gender and consumption of vegetables via perceived health, alcohol consumption via support from physicians, prior PCI and consumption of vegetables via results of care predicted perceived adherence to treatment indirectly. Healthcare professionals were an important factor in adherence to treatment, especially among women, those who were physically inactive, those with low vegetable consumption and those without relationships. In the future, multidisciplinary studies are needed to develop evidence-based interventions to support patient adherence to treatment.

ACK N OWLED G EM ENTS
We gratefully acknowledge the members of the PCICARE study Hospital Tampere. We also would like to thank nurses and all the patients who participated in the present study, and professor Pertti Töttö for statistical advice.

CO N FLI C T O F I NTE R E S T
No conflict of interest has been declared by the authors.

AUTH O R CO NTR I B UTI O N S
OK made substantial contributions to conception and design, or acquisition of data, or analysis and interpretation of data; OK, PK, TS, HM, PT, HK involved in drafting the manuscript or revising it critically for important intellectual content; Given final approval of the version to be published. Each author should have participated sufficiently in the work to take public responsibility for appropriate portions of the content; Agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

FU N D I N G I N FO R M ATI O N
The present study was supported by an educational grant from the Finnish Foundation of Cardiovascular Disease (16.4.2012) and Finnish Nursing Associations (6.6.2014).