Advertisement

Evaluating the Dimensionality and Reliability of the Thai Self-Care of Hypertension Inventory Version 2.0

Open AccessPublished:August 11, 2022DOI:https://doi.org/10.1016/j.anr.2022.08.002

      Summary

      Purpose

      Self-care is essential for hypertensive individuals to promote optimal health and illness treatment. We developed the Thai Self-Care of Hypertension Inventory (SC-HI) version 2.0 from the original US version using a multi-stage approach for cross-cultural adaptation. Scales previously studied outside a US context had different dimensions and factor solutions. Therefore, we examined the Thai SC-HI's factorial validity, construct validity, and internal reliability within a Thai context.

      Methods

      We administered a cross-sectional survey with hypertensive patients in 10 primary care settings, and conducted exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) on two sets of separate samples from each of five sites to examine the model's factorial validity and construct validity. We estimated scale reliability with Cronbach's alpha and McDonald's omega coefficients.

      Results

      Participants were predominantly female, older adults, with mean age 66 years (SD = 11.94; range 36–97 years). The self-care maintenance scale had three factors and demonstrated good fit when the error covariances were respecified. The two-factor self-care management scale had different factorial solutions compared to previous models. The CFA result showed good fit indices for the Thai, original US, and Brazilian models. The self-care confidence scale was unidimensional, with partially supported fit indices that improved after we respecified the error covariances. Reliability coefficients estimated by difference methods were nearly equal: slightly lower than desired for self-care maintenance (.68–.70) and inadequate for self-care management (.62–.65); self-care confidence reliability was adequate (.89–.90).

      Conclusion

      The Thai SC-HI has good psychometric characteristics and reflects the original instrument's theoretical basis.

      Keywords

      Introduction

      High blood pressure is the primary cause of cardiovascular disease, and is showing increased global prevalence. In Thailand, one in four adults has hypertension [
      • Aekplakorn W.
      • Hathaichanok P.
      • Kanittha T.
      • Warapone S.
      The 5th national health examination survey.
      ]. Better self-care that leads to appropriate lifestyle modifications is essential for controlling high blood pressure and reducing heart disease and stroke [
      • Burnier M.
      • Egan B.M.
      Adherence in hypertension.
      ,
      • Cornelissen V.A.
      • Smart N.A.
      Exercise training for blood pressure: a systematic review and meta-analysis.
      ,
      • Cook N.R.
      • Cutler J.A.
      • Obarzanek E.
      • Buring J.E.
      • Rexrode K.M.
      • Kumanyika S.K.
      • et al.
      Long term effects of dietary sodium reduction on cardiovascular disease outcomes: observational follow-up of the trials of hypertension prevention (TOHP).
      ]. As with other chronic illnesses, self-care is a surrogate hypertension health outcome, where a treatment regimen can promote and optimize overall health status [
      • Riegel B.
      • Jaarsma T.
      • Strömberg A.
      A middle-range theory of self-care of chronic illness.
      ]. Recommended hypertension self-care behaviors include: consuming optimal vegetable and fruit amounts, limiting sodium and fat, staying physically active, maintaining optimal body weight, reducing psychological stress, observing health condition changes, and managing high blood pressure symptoms [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ]. Although self-care's core concepts are universal, self-care behaviors are culturally embedded [
      • Riegel B.
      • Jaarsma T.
      • Strömberg A.
      A middle-range theory of self-care of chronic illness.
      ]. Because self-care is important to clinical outcomes, it is essential to have an effective measurement of hypertension self-care in diverse cultures.
      Self-care involves maintaining health and handling health condition changes through health promotion and illness management behaviors. The self-care instrument named Self-care of Hypertension Inventory (SC–HI) version 2.0 was developed in the US [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ]; therefore, cross-cultural adaptation was necessary to ensure its reliability and validity in a different cultural setting [
      • Phonphet C.
      • Suwanno J.
      • Thiamwong L.
      • Mayurapak C.
      • Ninla-Aesong P.
      Translation and cross-cultural adaptation of the self-care of hypertension inventory for Thais with hypertension.
      ,
      • Alsaqer K.
      • Bebis H.
      Cross-cultural adaptation, validity, and reliability of the Arabic version of the Self-care of Hypertension Inventory scale among older adults.
      ,
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ,
      • Zhao Q.
      • Guo Y.
      • Gu Y.
      • Yang L.
      Translation and cross-cultural adaptation of the Chinese version of the Self-care of Hypertension Inventory in older adults.
      ]. Two of three SC–HI scales (self-care maintenance and self-care management) measure self-care behaviors; the remaining self-care confidence scale measures self-care motivation [
      • Riegel B.
      • Jaarsma T.
      • Strömberg A.
      A middle-range theory of self-care of chronic illness.
      ,
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ]. The Thai SC–HI [
      • Phonphet C.
      • Suwanno J.
      • Thiamwong L.
      • Mayurapak C.
      • Ninla-Aesong P.
      Translation and cross-cultural adaptation of the self-care of hypertension inventory for Thais with hypertension.
      ] was developed using a global framework for the instrument's cross-cultural adaptation. Initial tests demonstrated that the Thai SC–HI was valid and reliable. Good relevance, clarity, simplicity, and ambiguity were reflected by the larger content validity indices supporting the original version [
      • Phonphet C.
      • Suwanno J.
      • Thiamwong L.
      • Mayurapak C.
      • Ninla-Aesong P.
      Translation and cross-cultural adaptation of the self-care of hypertension inventory for Thais with hypertension.
      ]. The Thai SC–HI's factorial validity, construct validity, and reliability required further testing because they are influenced by cultural context. Studies conducted outside the US that evaluated SC–HI versions showed that cross-cultural adaptations differed [
      • Alsaqer K.
      • Bebis H.
      Cross-cultural adaptation, validity, and reliability of the Arabic version of the Self-care of Hypertension Inventory scale among older adults.
      ,
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ,
      • Zhao Q.
      • Guo Y.
      • Gu Y.
      • Yang L.
      Translation and cross-cultural adaptation of the Chinese version of the Self-care of Hypertension Inventory in older adults.
      ]. For example, the self-care maintenance scale had a multidimensional structure [
      • Alsaqer K.
      • Bebis H.
      Cross-cultural adaptation, validity, and reliability of the Arabic version of the Self-care of Hypertension Inventory scale among older adults.
      ,
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ,
      • Zhao Q.
      • Guo Y.
      • Gu Y.
      • Yang L.
      Translation and cross-cultural adaptation of the Chinese version of the Self-care of Hypertension Inventory in older adults.
      ], different items were allocated to the self-care management scale's autonomous and consultative dimensions [
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ], and item correlations differed in the self-care confidence scale [
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ]. The cross-culturally developed SC–HI's dimensionality and psychometric properties showed that it is valid and reliable, has cultural diversity, and generally verified and re-conceptualized [
      • Alsaqer K.
      • Bebis H.
      Cross-cultural adaptation, validity, and reliability of the Arabic version of the Self-care of Hypertension Inventory scale among older adults.
      ,
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ,
      • Zhao Q.
      • Guo Y.
      • Gu Y.
      • Yang L.
      Translation and cross-cultural adaptation of the Chinese version of the Self-care of Hypertension Inventory in older adults.
      ].
      Few cross-cultural SC–HI's have been adapted within an Asian context [
      • Phonphet C.
      • Suwanno J.
      • Thiamwong L.
      • Mayurapak C.
      • Ninla-Aesong P.
      Translation and cross-cultural adaptation of the self-care of hypertension inventory for Thais with hypertension.
      ,
      • Zhao Q.
      • Guo Y.
      • Gu Y.
      • Yang L.
      Translation and cross-cultural adaptation of the Chinese version of the Self-care of Hypertension Inventory in older adults.
      ], and these have only preliminary testing. Full psychometric property evaluation has not been conducted. The SC–HI is a theory-based instrument developed to assess naturalistic chronic illness self-care [
      • Riegel B.
      • Jaarsma T.
      • Strömberg A.
      A middle-range theory of self-care of chronic illness.
      ,
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ]. Hence, it measures the generic concept of self-care that reflects health-promoting and illness-related behaviors employed to maintain health, manage symptoms, and enhance confidence and efficacy. Each of the three scales includes items that measure specific self-care actions followed within a hypertension treatment regimen. Although self-care is a global concept and self-care behavior is disease specific (hypertension in this case), behavioral practices (e.g., dietary patterns, consumed foods, health care contacts, stress management) vary by cultural background [
      • Phonphet C.
      • Suwanno J.
      • Thiamwong L.
      • Mayurapak C.
      • Ninla-Aesong P.
      Translation and cross-cultural adaptation of the self-care of hypertension inventory for Thais with hypertension.
      ,
      • Zhao Q.
      • Guo Y.
      • Gu Y.
      • Yang L.
      Translation and cross-cultural adaptation of the Chinese version of the Self-care of Hypertension Inventory in older adults.
      ]. Thus, a cross-cultural adaptation instrument is needed to test whether it is valid and reliable within the cultural context of where it is used. Previous validated and reliable hypertension self-care instruments were developed, tested, or used in Asia countries [
      • Ma Y.
      • Cheng H.Y.
      • Sit J.W.H.
      • Chien W.T.
      Psychometric evaluation of the Chinese version of hypertension self-care profile.
      ,
      • Wee S.Y.
      • Salim H.
      • Mawardi M.
      • Koh Y.L.E.
      • Ali H.
      • Shariff Ghazali S.
      • et al.
      Comparing and determining factors associated with hypertension self-care profiles of patients in two multi-ethnic Asian countries: cross-sectional studies between two study populations.
      ,
      • Lee J.E.
      • Han H.R.
      • Song H.
      • Kim J.
      • Kim K.B.
      • Ryu J.P.
      • et al.
      Correlates of self-care behaviors for managing hypertension among Korean Americans: a questionnaire survey.
      ]. However, most were used to determine health maintenance behaviors that shared common self-care actions, and they were not comprehensive. A theory-based instrument is needed to evaluate overall self-care tasks that manage both stable illness stages and symptom changes in chronic illness, including hypertension [
      • Riegel B.
      • Barbaranelli C.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • Miller J.L.
      • et al.
      Development and initial testing of the self-care of chronic illness inventory.
      ].
      This study tested the SC-HI's psychometric properties in Thai individuals with hypertension. First, we conducted an exploratory analysis (EFA) to assess the three scales' factorial validity. Next, validate the US and Brazilian versions of the self-care management scale with Thai data using confirmatory factor analysis (CFA) to test construct validity [
      • Osborne J.W.
      Best practice in exploratory factor analysis.
      ]. A modification model that fit the Thai data was respecified for all scales. We also evaluated discriminant and convergent validity. Cronbach's alpha coefficients and other criteria for multidimensional scales were used to test each scale's internal consistency reliability. We used composite reliability [
      • Raykov T.
      Handbook of structural equation modeling. Reprint ed..
      ] or McDonald's omega [
      • McDonald R.P.
      Test theory: a unified treatment.
      ] to evaluate multidimensional scales' reliability [
      • Riegel B.
      • Barbaranelli C.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • Miller J.L.
      • et al.
      Development and initial testing of the self-care of chronic illness inventory.
      ,
      • Trizano-Hermosilla I.
      • Gálvez-Nieto J.L.
      • Alvarado J.M.
      • Saiz J.L.
      • Salvo-Garrido S.
      Reliability estimation in multidimensional scales: comparing the bias of six estimators in measures with a bifactor structure.
      ,
      • Barbaranelli C.
      • Lee C.S.
      • Vellone E.
      • Riegel B.
      Dimensionality and reliability of the self-care of heart failure index scales: further evidence from confirmatory factor analysis.
      ] because these methods were less biased [
      • Trizano-Hermosilla I.
      • Gálvez-Nieto J.L.
      • Alvarado J.M.
      • Saiz J.L.
      • Salvo-Garrido S.
      Reliability estimation in multidimensional scales: comparing the bias of six estimators in measures with a bifactor structure.
      ]. We hypothesized that the SC–HI Thai version was valid and reliable, with adequate goodness of fit indices, discriminant validity, convergent validity, and the internal coherence.

      Theoretical framework

      The SC–HI is a theory based and disease specific measure of hypertension self-care that includes maintenance, management, and confidence components. It was derived from the middle-range self-care for chronic illness theory [
      • Riegel B.
      • Jaarsma T.
      • Strömberg A.
      A middle-range theory of self-care of chronic illness.
      ], and the situation-specific heart failure self-care theory [
      • Riegel B.
      • Dickson V.V.
      • Faulkner K.M.
      The situation-specific theory of heart failure self-care: revised and updated.
      ]. The theories describe self-care as a realistic decision-making process that promotes health behavior and disease treatment. Self-care maintenance includes behaviors used to maintain physical and emotional stability. Self-care management involves responding to symptoms. Both self-care maintenance and management are motivated by self-confidence [
      • Riegel B.
      • Jaarsma T.
      • Strömberg A.
      A middle-range theory of self-care of chronic illness.
      ,
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ].

      Methods

      Study design

      A multi-site, cross-sectional study design was conducted from May 2017 to April 2019 to evaluate the scales' psychometric properties. We investigated cardiometabolic risk factors, self-care, and hypertension outcomes as described in Identification of Complex Care Needed in Patients with Hypertension Treated at Primary Care (ICNHT) [
      • Phonphet C.
      • Suwanno J.
      • Thiamwong L.
      • Mayurapak C.
      • Ninla-Aesong P.
      Translation and cross-cultural adaptation of the self-care of hypertension inventory for Thais with hypertension.
      ].

      Ethical considerations

      Approval was obtained from the University Ethics Board Committee (code number: 59/075). The study adhered to standards delineated in the Declaration of Helsinki.

      Participants

      Participants were recruited from ten health promotion hospitals (HPH) in one Southern Thailand province. Inclusion criteria: individuals with hypertension who are treated with any antihypertensive medication for at least six months. Pregnant women with hypertension and individuals who were unable to communicate were excluded. EFA was conducted with data from the first set of participants from five sites to test the dimensionality and factor loading on each of the scales' dimensions. CFA was conducted on a different set of samples from five other sites to evaluate validity (construct, discriminant, and convergent) and internal coherence. Generally, a sample size of 200 is required for psychometric testing [
      • Anthoine E.
      • Moret L.
      • Regnault A.
      • Sebille V.
      • Hardouin J.B.
      Sample size used to validate a scale: a review of publications on newly-developed patient reported outcomes measures.
      ]. However, for this analysis we enrolled the entire sample from the target settings to allow cross-validation [
      • Riegel B.
      • Barbaranelli C.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • Miller J.L.
      • et al.
      Development and initial testing of the self-care of chronic illness inventory.
      ].
      HPHs are Thailand's first level health care facilities; they provide primary care in the subdistrict area and are covered by the district hospital. Our setting is among Thailand's largest provinces, reaching 1.6 million people, with four geographical areas: provincial or central district, hill-side, sea-side, and field or catchment basins. Each HPH treats 200 to 500 patients. Participants were recruited using a multi-stage cluster sampling method. Seven districts and one center were randomly selected based on geographical area. Two HPHs were chosen using simple random sampling, which yielded eight rural and two urban settings, with ten HPHs in total. Four rural and one urban setting were assigned to the EFA and CFA groups. Approximately 100 to 150 participants were selected by simple random sampling from each HPH. The target sample was stratified by 10-year age groups (i.e., <40; 40–49; … 80 and older). All men were recruited because of the small number of men across all age groups.

      Thai SC–HI (version 2.0) instrument

      The Thai SC–HI's content validity, face validity, test–retest reliability, and interobserver reliability was presented elsewhere [
      • Phonphet C.
      • Suwanno J.
      • Thiamwong L.
      • Mayurapak C.
      • Ninla-Aesong P.
      Translation and cross-cultural adaptation of the self-care of hypertension inventory for Thais with hypertension.
      ]. In brief, following the original US version [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ], we used a multi-stage approach for cross-cultural adaptation (https://self-care-measures.com). The SC–HI's 23 items comprise three scales: self-care maintenance (11 items: scmt 01–11 of the original version), self-care management (6 items: scmn 12–17 or items 13–18 of the original version), and self-care confidence (6 items: scc 18–23 or items 19–24 of the original version). The items are rated on Likert-type scales from 0 to 4 or 1 to 4. The self-care management scale is administered to individuals who experience hypertension-related symptoms or uncontrolled blood pressure within four weeks. Response choices for all items are standardized for scores from 0 to 100, where higher scores indicate better self-care. The Thai SC–HI's item-level and scale-level content validity index was 1.00. The item-level intraclass correlation coefficients (ICCs) ranged from .97 to 1.00 for interobserver agreement and .95 to 1.00 for test–retest. The overall scale and three distinct scales' interobserver ICCs were .99. The test–retest ICCs were .99 for the total scale, and ranged from .97 to .99 for the three individual scales [
      • Phonphet C.
      • Suwanno J.
      • Thiamwong L.
      • Mayurapak C.
      • Ninla-Aesong P.
      Translation and cross-cultural adaptation of the self-care of hypertension inventory for Thais with hypertension.
      ].
      In the original version, self-care management and self-care confidence were one-factor models, whereas self-care management was a two-factor model [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ]. We found the following differences in the items allocated to self-care management components across the two existing models [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ,
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ]: autonomous (US: scmn 14, 15, and 17; Brazilian: scmn 12, 15, and 17); and consultative (US: scmn 12, 13, and 16; Brazilian; scmn 13, 14, and 16). Thus, we conducted CFA on all Self-Care Management models where the Thai model EFA factorial differed from the previous models.

      Data collection

      Data were collected by five graduated nursing students. All attended the 72-hour clinical-based sessions on cardiometabolic risk and self-care assessment research protocols. The SC–HI data were collected during a face-to-face interview at the HPH or the participant's home, according to their preference. Participants were asked to complete the questionnaire by themselves or have the research assistant read it for them. Participants' sociodemographic and clinical data were obtained from electronic health records and physical examination, including: an average of two blood pressure measurements, body mass index, waist circumference, history of diabetes, history of dyslipidemia, self-reported daily or occasional smoking or alcohol consumption within the last six months. The sociodemographic and clinical data were record in a data collection form.

      Data analysis

      Descriptive statistics, EFA, and internal consistency reliability were analyzed with IBM SPSS Statistics, version 28.0; CFA was conducted with AMOS version 24.0. We checked for multivariate outliers using Mahalanobis distance and excluded them from all subsequent analyses [
      • Tabachnick B.G.
      • Fidell L.S.
      Using multivariate statistics.
      ]. Descriptive statistics included percentage (%), number, median (interquartile rank [IQR]) and mean (standard deviation [SD]), and assessed individual item skewness and kurtosis to identify the estimator [
      • Osborne J.W.
      Best practice in exploratory factor analysis.
      ,
      • Muthén B.
      • Kaplan D.A.
      Comparison of some methodologies for the factor analysis of non-normal Likert variables.
      ]. Each scale's scores were standardized [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ] on a scale of 0 to 100.
      EFA was conducted before CFA, using the first data set to test the three separate scales' dimension structure. A Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy of .60 or greater and a significant Bartlett's test of sphericity showed that the correlation matrix was suitable for factor analysis [
      • Kaiser H.F.
      An index of factorial simplicity.
      ]. EFA was first performed unrotated, using principal axes factoring with non-normality data and maximum likelihood with multivariate normality [
      • Osborne J.W.
      Best practice in exploratory factor analysis.
      ]. Factors with eigenvalues greater than 1.00 were retained [
      • Osborne J.W.
      Best practice in exploratory factor analysis.
      ]. We performed EFA with varimax rotation and enforced one-, two- and three-factor solutions to evaluate the scale's factorial validity.
      CFA was conducted on the second data set to validate the three scales' construct validity. We used the robust maximum likelihood method to estimate the parameters because various non-normalized items and factor loadings above .30 were acceptable [
      • Riegel B.
      • Barbaranelli C.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • Miller J.L.
      • et al.
      Development and initial testing of the self-care of chronic illness inventory.
      ,
      • Riegel B.
      • Barbaranelli C.
      • Carlson B.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • et al.
      Psychometric testing of the revised self-care of heart failure index.
      ]. As done for the original version [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ] and other self-care instruments [
      • Riegel B.
      • Barbaranelli C.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • Miller J.L.
      • et al.
      Development and initial testing of the self-care of chronic illness inventory.
      ,
      • Riegel B.
      • Barbaranelli C.
      • Carlson B.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • et al.
      Psychometric testing of the revised self-care of heart failure index.
      ], we used numerous goodness of fit indices to analyze the model fit: the Tucker–Lewis index (TLI), comparative fit index (CFI), standardized root mean square residual (SRMR), and root mean square error of approximation (RMSEA). CFI and TLI values above .95 indicate good model fit, and .90–.95 values indicate acceptable fit [
      • Hu L.
      • Bentler P.M.
      Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives.
      ]. However, the TLI is a non-normed incremental fit index; therefore, the values can fall outside the 0–1.00 range [
      • Hu L.
      • Bentler P.M.
      Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives.
      ]. We used these values to compare the desirable model with a null model. We used RMSEA to estimate model fit, where a value of .05 or less indicates good fit; .05–.08, moderate fit; .10 or above, poor fit [
      • Browne M.W.
      • Cudek R.
      Alternative ways of assessing model fit.
      ]. We used SRMR to measure absolute fit, where a value of .08 or less indicates good fit, and .10 or above indicates poor fit [
      • Browne M.W.
      • Cudek R.
      Alternative ways of assessing model fit.
      ]. The Chi-square likelihood ratio is sensitive to sample size; hence, we reported the traditional Chi-square test, but did not use it to interpret model fit.
      We used the CFA data set to evaluate others characteristic of construct validity, discriminant validity and convergent validity [

      Krabbe PFM. The measurement of health and health status: concepts, methods and applications from a multidisciplinary perspective [Internet], Version 1.0 [revised 2016 Oct 7; cited 2022 May 18]. Available from: https://doi.org/10.1016/B978-0-12-801504-9.00007-6.

      ]. Discriminant validity is the extent to which the factors are distinct and not highly correlated [
      • Hubley A.M.
      Discriminant validity.
      ]. A new criterion for assessing discriminant validity, the heterotriat-monotrait (HTMT) ratio was used for the multidimensional scales, using a criterion of < .85 to indicate discriminant validity [
      • Henseler J.
      • Ringle C.M.
      • Sarstedt M.
      A new criterion for assessing discriminant validity in variance-based structural equation modeling.
      ]. For the HTMT technique, the scale score correlation is converted to an error-adjusted correlation using parallel reliability [
      • Ronkko M.
      • Cho E.
      An updated guideline for assessing discriminant validity.
      ]. Also, convergent validity provides supporting evidence for construct validity. We considered the correlation between factors within the scale to measure the same construct. We analyzed Pearson's coefficient correlation (r) by using the standardized scores of the multidimensional scales. A positive and significant correlation between two factors within the same scale was considered evidence of convergent validity [

      Krabbe PFM. The measurement of health and health status: concepts, methods and applications from a multidisciplinary perspective [Internet], Version 1.0 [revised 2016 Oct 7; cited 2022 May 18]. Available from: https://doi.org/10.1016/B978-0-12-801504-9.00007-6.

      ].
      Internal consistency was also tested using the CFA data set. We estimated scale reliability with Cronbach's alpha coefficients and composite reliability [
      • Raykov T.
      Handbook of structural equation modeling. Reprint ed..
      ] or McDonald's omega coefficients [
      • McDonald R.P.
      Test theory: a unified treatment.
      ], with .70 or above considered acceptable [
      • Bagozzi R.P.
      • Yi Y.
      Specification, evaluation, and interpretation of structural equation models.
      ]. Similar to the alpha coefficient, the omega coefficient evaluates scale-level reliability. However, omega has an advantage with multidimensional scales [
      • Trizano-Hermosilla I.
      • Gálvez-Nieto J.L.
      • Alvarado J.M.
      • Saiz J.L.
      • Salvo-Garrido S.
      Reliability estimation in multidimensional scales: comparing the bias of six estimators in measures with a bifactor structure.
      ]. We calculated omega from each scale's total items, similar to the relevant self-care instruments, because self-care interpretation is meaningful by the scale-level [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ,
      • Riegel B.
      • Barbaranelli C.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • Miller J.L.
      • et al.
      Development and initial testing of the self-care of chronic illness inventory.
      ]. Item discrimination was estimated using item-total corrected correlation coefficients, with .30 or above considered acceptable [
      • Nunnally J.
      • Bernstein I.H.
      Psychometric theory. 3rd rev. ed..
      ].

      Results

      Participant characteristics

      We enrolled 1,262 adults from ten primary care settings; the final sample included 1,214 individuals after excluding 48 outlier cases. Participants were predominantly female, older adults, with and mean age of 66 years (range 36–97 years), and were socioeconomically well off (Table 1). Rates of each clinical risk factor, including uncontrolled blood pressure, were 6.8%–60.7%, and 81.5% had been treated for hypertension for over two years. The standardized mean scores (± SD) of each scale were 54.84 (SD = 14.51), 52.19 (SD = 13.57), and 50.93 (SD = 19.38) for self-care maintenance, self-care management, and self-care confidence, respectively. EFA and CFA showed different rates in literacy, known and treated diabetes, known and treated dyslipidemia, and abdominal obesity, while the three scales' self-care scores and total SC–HI scores did not differ.
      Table 1Sociodemographic and Clinical Characteristics of the Participants.
      CharacteristicsOverall sampleEFA sampleCFA samplep
      Participant, n (%)1,214 (100)640 (52.7)574 (47.3)
      Sociodemographic
      Women889 (73.2)477 (74.1)412 (72.3).483
      Age, mean (SD)66.12 (11.94)65.89 (11.46)66.39 (12.47).466
      Age, range in years36–9737–9536–97
      Age ≥ 65 years old672 (55.4)345 (53.6)327 (57.4).184
      Living with spouse or family members984 (81.1)516 (80.1)468 (82.1).379
      Primary education1,102 (90.8)578 (89.8)524 (91.9).191
      Literate1,120 (92.3)616 (95.7)504 (88.4)<.001
      Agriculture1,117 (92.0)593 (92.1)524 (91.9).923
      Working or employed645 (53.1)359 (55.7)286 (50.2).052
      Family income sufficiency900 (74.1)467 (72.5)433 (76.0).171
      Clinical characteristics
      Systolic blood pressure ≥ 140 mmHg291 (24.0)168 (26.1)123 (21.6).066
      Diastolic blood pressure ≥ 90 mmHg98 (8.1)56 (8.7)72 (7.4).397
      Known and treated diabetes371 (30.6)252 (39.1)119 (20.9)<.001
      Known and treated dyslipidemia639 (52.6)405 (62.9)234 (41.1)<.001
      Body mass index ≥ 25.0 kg/m2532 (43.8)295 (45.8)237 (41.6).128
      Abdominal obesity737 (60.7)422 (65.5)315 (55.3)<.001
      Currently smoking132 (10.9)74 (11.5)58 (10.2).463
      Currently drinking alcohol83 (6.8)46 (55.4)37 (6.5).653
      Number of hypertension medication use, median (IQR)2 (1, 3)2 (1, 3)2 (1, 3).295
      Duration of hypertension >2 years989 (81.5)524 (81.4)465 (81.6).924
      Self-care of hypertension
      Self-care maintenance scale54.84 (14.51)53.46 (13.90)54.46 (15.03).367
      Self-care management scale52.19 (13.57)51.70 (13.90)52.69 (13.20).370
      Self-care confidence scale50.93 (19.38)51.41 (19.66)50.39 (19.06).358
      Self-care of hypertension total54.81 (11.33)54.67 (11.27)54.94 (11.41).770
      Note. Values are n (%), mean (standard deviation [SD]), and median (interquartile rank [IQR]).
      Abbreviation: CFA = confirmatory factor analysis; DBP = diastolic blood pressure; EFA = exploratory factor analysis.

      Item response

      Table 2 shows item response means and distribution. All items had scores above the lowest, except one Self-Care Management scale item (scmt 01); two items had below average mean scores (items: scmt 01 and scmt 12); and two items had non-normal distributions (items: scmt 04 and scmt 07).
      Table 2Self-care of Hypertension Item Descriptive Analysis.
      ItemsMean (SD)SkewnessKurtosis
      Self-care maintenance scale
      Scmt 01. Check your blood pressure.1.70 (0.85)1.110.53
      Scmt 02. Eat a variety of vegetables, fruits …2.70 (0.79)0.07−0.66
      Scmt 03. Exert on doing daily busy activity, …2.84 (0.91)−0.23−0.89
      Scmt 04. Attend hospital for routine follow-up …3.77 (0.51)−2.184.09
      Scmt 05. Eat a less salty foodstuff.2.67 (0.98)−0.05−1.07
      Scmt 06. Exercise for at least a half-hour.2.52 (1.01)0.08−1.11
      Scmt 07. Take medicines as prescribed.3.82 (0.45)−2.526.13
      Scmt 08. Selected less salty food choices …2.22 (0.93)0.45−0.61
      Scmt 09. Use a system to help you remember your medicines …2.21 (1.28)0.36−1.44
      Scmt 10. Avoiding high-fatty foodstuff.2.39 (0.89)0.22−0.69
      Scmt 11. Try to lower your weight …2.27 (1.01)0.28−1.01
      Self-care management scale
      Scmn 12. How quickly did you recognize that your blood pressure was up …1.79 (0.72)0.851.01
      Scmn 13. Reduce the salt or salty recipes in your meal …2.86 (0.78)−0.35−0.23
      Scmn 14. Mindful relaxation, be aware of stress …2.89 (0.71)−0.370.13
      Scmn 15. Be strict on taking your blood pressure-lowering medicines more regularly.3.49 (0.67)−1.040.30
      Scmn 16. Contact your healthcare provider …2.39 (1.06)0.04−1.26
      Scmn 17. How sure were you that the action helped …2.01 (0.69)0.041.56
      Self-care confidence scale
      Scc 18. Control your blood pressure.2.49 (0.71)0.22−0.24
      Scc 19. Follow your hypertension treatment regimen.2.59 (0.73)0.19−0.39
      Scc 20. Recognize when your health is out of the ordinary.2.51 (0.71)0.11−0.26
      Scc 21. Evaluate whether either your blood pressure was up …2.49 (0.70)0.03−0.23
      Scc 22. Take action that will control your blood pressure.2.53 (0.71)0.14−0.27
      Scc 23. Evaluate how well a self-care action works.2.57 (0.70)0.10−0.29

      Factorial validity

      The scale structure was evaluated based on EFA and the rotated factor loadings matrix in 640 samples, as shown in Table 3. The self-care maintenance, self-care management, and self-care confidence scales' KMO values were .68, .62, and .88, respectively. Bartlett's test of sphericity yielded p < .001 for all scales, indicating factor analysis suitability. We used a principal axes factor method [
      • Osborne J.W.
      Best practice in exploratory factor analysis.
      ] with no rotation and exacted three factors for self-care maintenance, and two factors for self-care management. We used the maximum likelihood method [
      • Osborne J.W.
      Best practice in exploratory factor analysis.
      ] with no rotation and exacted one-dimension for self-care confidence. Each scale's eigenvalues were: self-care maintenance (factor 1 = 2.59, factor 2 = 1.63, and factor 3 = 1.31); self-care management (factor 1 = 2.04 and factor 2 = 1.27); and self-care confidence (one-factor = 3.97). One-factor and two-factor models for self-care maintenance revealed eigenvalues as follows: one-factor (factor 1 = 1.82); and two-factor (factor 1 = 1.84 and factor 2 = 1.07).
      Table 3Factors Loading Matrix, Communalities Values and Eigenvalue Values of the Rotated Model of Self-Care Maintenance, Self-Care Management, and Self-Care Confidence Scales.
      One-factorTwo-factorThree-factor
      Factor 1h2Factor 1Factor 2h2Factor 1Factor 2Factor 3h2
      Self-care maintenance scale (N = 640)
      Scmt 01..203.041.206.001.043.088.200.024.048
      Scmt 02..395.156.409.090.175.229.335.051.168
      Scmt 03..425.180.457.167.237.065.683.046.473
      Scmt 04..056.003.011.668.446.029.015.729.533
      Scmt 05..387.150.377.078.148.471.049.003.224
      Scmt 06..483.233.515.150.288.102.768.024.602
      Scmt 07..076.006.034.723.525.073.029.734.545
      Scmt 08..546.299.535.067.291.566.179.002.352
      Scmt 09..445.198.433.091.196.484.105.024.246
      Scmt 10..458.210.443.155.221.537.076.074.300
      Scmt 11..592.350.580.127.352.508.274.094.342
      Self-care management scale (N = 298)
      Scmn 12..108.336.124
      Scmn 13..920.077.853
      Scmn 14..660.199.476
      Scmn 15..206.118.056
      Scmn 16..451.035.205
      Scmn 17..025.904.817
      Self-care confidence scale (N = 640)
      Scc 18..830.689
      Scc 19..789.622
      Scc 20..810.656
      Scc 21..794.630
      Scc 22..837.701
      Scc 23..823.677
      Extraction Eigenvalues values and percentage (%) of variance by factors (Rotation sums of squared loading)
      Self-care maintenanceSelf-care managementSelf-care confidence
      Factor 1Factor 2Factor 3Factor 1Factor 2Factor 1
      Eigenvalues2.591.631.312.041.273.97
      % of variance12.7712.159.9225.6816.4966.25
      Cumulative %12.7724.9234.8425.6842.1766.25
      Note. h2 are the extraction communalities values.

      Self-care maintenance scale

      The principal axes factor method [
      • Osborne J.W.
      Best practice in exploratory factor analysis.
      ] with varimax rotation exacted three factors with the following total variance: Factor 1 comprised five items and explained 12.8% of the variance; Factor 2 comprised four items and explained 12.2% of the variance; and Factor 3 comprised two items and explained 9.9% of the variance. Factor loadings for each item in Factors 1, 2, and 3 ranged from .471 to .566, .200 to .768, and .729 to .734, respectively, explaining 34.8% of the total variance. The extraction communalities values were .048 to .602, and one item (scmt 01) with a rotated factor loading matrix was below .30. A two-factor model shows the factors' explanation power relative to the total variance: Factor 1 comprised nine items and explained 16.6% of the variance, and Factor 2 comprised two items and explained 9.9% of the variance. Factor loadings for each item in factors 1 and 2 ranged from .206 to .580 and .668 to .723, respectively, explaining 26.6% of the total variance. The extraction communalities values were .043 to .525, and one item (scmt 01) with a rotated factor loading matrix was below .30. Finally, when we fixed the number of factor to a one-factor model as of the original US model, the factor loading was .056 to .592 and explained 16.6% of the total variance. The extraction communalities values were .003 to .350, and three items (scmt 01, scmt 04, and scmt 07) with a rotated factor loading matrix were below .30.

      Self-care management scale

      The principal axes factor method [
      • Osborne J.W.
      Best practice in exploratory factor analysis.
      ] with varimax rotation exacted two dimensions: Factor 1 comprised four items and explained 25.7% of the variance; Factor 2 comprised two items and explained 16.5% of the variance. Factor loadings of each item in Factors 1 and 2 ranged from .206 to .920 and .336 to .904, respectively, explaining 42.2% of the total variance. The extraction communalities values were .056 to .853, and one item (scmn 15) with a rotated factor loading matrix was below .30.

      Self-care confidence scale

      The maximum likelihood method [
      • Osborne J.W.
      Best practice in exploratory factor analysis.
      ] with varimax rotation exacted a one-factor model. The explanatory power of the factors relative to the total variance of 66.3% had factor loadings ranging from .789 to .837. The extraction communalities values were .622 to .701.

      Construct validity

      Table 4 and Figure 1, Figure 2, Figure 3 show the CFA results. We evaluated the initial model fit indices with 547 participants, and respecified models for self-care maintenance (Figure 1), self-care management (Figure 2A–C), and Self-Care Confidence (Figure 3).
      Table 4Model Fit Summary for the Thai Self-Care of Hypertension Inventory.
      Chi-squaredfpCFITLIRMSEA (90% CI)SRMR
      Initial models
      Self-care maintenance scale
      One-factor465.7944<.001.594.492.129 (.119–.140).063
      Two-factor249.3443<.001.801.746.092 (.081–.103).060
      Three-factor125.2541<.001.919.891.060 (.048–.072).044
      Self-care management scale
      Thai model3.838.8721.0001.037.000 (.000–.034).010
      US model6.128.6341.0001.016.000 (.000–.056).012
      Brazilian model5.948.6541.0001.018.000 (.000–.055).012
      Self-care confidence scale
      One-factor96.149<.001.955.925.130 (.107–.154).019
      Respecified models
      Self-care maintenance scale
      Two-factor92.0839<.001.950.928.049 (.036–.062).037
      Three-factor98.9439<.001.942.919.052 (.039–.065).038
      Self-care confidence scale
      One-factor22.487.002.992.983.062 (.036–.093).008
      Figure 1
      Figure 1The standardized estimates of self-care maintenance scale with confirmatory factor analysis. Note. HPB = health promoting behaviors; IRB = illness related behaviors; MCB = medical care behaviors; scmt = self-care management; scmt 01 = item number 01. The numbers near the one headed arrows are factor loading coefficients; the numbers near the two-headed arrows are correlation coefficients.
      Figure 2
      Figure 2The standardized estimates of self-care management scale with confirmatory factor analysis. (A). Bidimensional self-care management Thai model. (B). Bidimensional self-care management US model. (C). Bidimensional self-care management Brazilian model. Note. scmn = self-care management; scmn 13 = item number 13. The numbers near the one headed arrows are factor loading coefficients.
      Figure 3
      Figure 3The standardized estimates self-care confidence scale with confirmatory factor analysis. Note. scc = self-care confidence; scc18 = item number 18. The numbers near the one headed arrows are factor loading coefficients; the numbers near the two-headed arrows are correlation coefficients.

      Self-care maintenance scale

      Three models of self-care management were analyzed. We first specified a one-factor confirmatory model per the original version [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ]. The model's goodness of fit indices was unacceptable: χ2/df (44, N = 574) = 465.79, p < .001, CFI = .594, TLI = .492, RMSEA = .129 (90% CI = .119–.140), SRMR = .063). Three items had factor loadings <.03 (items: scmt 04 “Attend hospital for routine follow-up”; scmt 07 “Take medicines as prescribed”; and scmt 09 “Use a system to help you remember your medicines.” Second, since self-care maintenance includes health promoting behavior and illness related behavior dimensions as relevant self-care measures [
      • Riegel B.
      • Barbaranelli C.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • Miller J.L.
      • et al.
      Development and initial testing of the self-care of chronic illness inventory.
      ], we specified a two-factor model. The EFA results allocated nine items from health promoting behavior and two items from illness related behavior dimensions. The model's goodness of fit indices was partially supported: χ2/df (43, N = 574) = 249.34, p < .001, CFI = .801, TLI = .746, RMSEA = .092 (90% CI = .081–.103), SRMR = .060). One item (scmt 09) had a factor loading <.03. We respecified a two-factor model and the modified indices estimated three error covariances. One variance was between items scmt 03 “exert on doing daily busy activity” and scmt 06 “exercise for at least a half-hour.” Items scmt 05 “eat less salty foodstuff” and scmt 08 “selected less salty food choices,” and scmt 10 “Avoiding high-fat foodstuff” covaried. Items scmt 08 and scmt 10 also covaried. Model fit was acceptable with partially supported goodness fit indices when respecified using these error covariances: χ2/df (39, N = 574) = 92.08, p < .001, CFI = .950, TLI = .928, RMSEA = .049 (90% CI = .036–.062), SRMR = .037). All factor loadings were significantly positive. No items had factor loadings below .30. The two dimensions were positively correlated at .21.
      Finally, we tested a three-factor model. Items allocated to each factor followed the EFA results. The model's goodness of fit indices was good: χ2/df (41, N = 574) = 125.25, p < .001, CFI = .919, TLI = .891, RMSEA = .060 (90% CI = .048–.072), SRMR = .044). One item (scmt 01 “check your blood pressure”) had a factor loading <.03. We also respecified the three-factor model (Figure 1). The modification indices estimated two error covariances between items scmt 11 “try to lower your weight” and scmt 02 “eat a variety of vegetables, fruits, and grains,” and between scmt 11 and scmt 06 “exercise for at least a half-hour.” The final model's goodness of fit indices was improved: χ2/df (39, N = 574) = 98.94, p < .001, CFI = .942, TLI = .919, RMSEA = .052 (90% CI = .039–.065), SRMR = .038). Still, one item, scmt 01 had a factor loading <.03. Two items (scmt 04 and scmt 07) allocated in the third factor reflected “medical care behaviors.” The health promoting behaviors and illness related behaviors dimensions comprised five and four items, respectively. A positive correlation among each pair's dimension was observed at .12 to .49.

      Self-care management scale

      The recent Thai, and the previous US [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ] and Brazilian [
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ] self-care management models had different items belonging to autonomous and consultative dimensions. We conducted CFA on these three models. All models' goodness of fit indices suggested a perfect fit (Table 4, Figure 2A–C). Thai model: χ2/df (8, N = 305) = 3.83, p = .872, CFI = 1.000, TLI = 1.037, RMSEA = .000 (90% CI = .000–.034), SRMR = .010); US model: χ2/df (8, N = 305) = 6.12, p = .634, CFI = 1.000, TLI = 1.016, RMSEA = .000 (90% CI = .000–.056), SRMR = .012); and Brazilian model: χ2/df (8, N = 305) = 5.94, p = .654, CFI = 1.000, TLI = 1.018, RMSEA = .000 (90% CI = .000–.055), SRMR = .012). All factor loadings for each model were positive and significant. Item scmn 12 from the US [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ] and Brazilian [
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ] models had factor loading below .30. The two dimensions were positively correlated at .64, .97, and .91 for the Thai, US, and Brazilian models, respectively.

      Self-care confidence scale

      When we tested the one-factor CFA from the original model, the goodness of fit indices were adequate but only partially supportive (Table 4): χ2/df (9, N = 574) = 96.14, p < .001, CFI = .955, TLI = .925, RMSEA = .130 (90% CI = .107–.154), SRMR = .019). The modification indices revealed three error covariances; one between items scc18 “control your blood pressure” and scc19 “follow your hypertension treatment regimen”; another between items scc 18 and scc 20 “recognize when your health is out of the ordinary.” Model fit was good when the model was respecified with these error covariances (Table 4 and Figure 3): χ2/df (7, N = 574) = 22.48, p = .002, CFI = .992, TLI = .983, RMSEA = .062 (90% CI = .036–.093), SRMR = .008). All factor loadings were positive and significant. No items had factor loadings below .30.

      Discriminant validity

      We examined discriminant validity in the self-care maintenance scale three-factor model by calculating HTMT ratios [
      • Henseler J.
      • Ringle C.M.
      • Sarstedt M.
      A new criterion for assessing discriminant validity in variance-based structural equation modeling.
      ] among the three factors, which were less than .85, indicating acceptable discriminant validity. The HTMT ratios were .54 between health promoting behavior and medical care behavior, .12 between health promoting behavior and medical care behavior, and .23 between illness related behavior and medical care behavior. Self-care management discriminant validity was also supported by HTMT ratios [
      • Henseler J.
      • Ringle C.M.
      • Sarstedt M.
      A new criterion for assessing discriminant validity in variance-based structural equation modeling.
      ] lower than .85. The HTMT ratio within the Thai two-factor model was .65 between consultative behavior and autonomous behavior.

      Convergent validity

      We evaluated convergent validity within the self-care maintenance scale three-factor model and found positive and significant correlations between pair dimensions [

      Krabbe PFM. The measurement of health and health status: concepts, methods and applications from a multidisciplinary perspective [Internet], Version 1.0 [revised 2016 Oct 7; cited 2022 May 18]. Available from: https://doi.org/10.1016/B978-0-12-801504-9.00007-6.

      ]. Pearson correlation coefficients were 0.38 (p = .01) between health promoting behavior and illness related behavior, .11 (p < .05) between health promoting behavior and medical care behavior, and .15 (p = .01) between illness related behavior and medical care behavior. Also, we observed a positive and significant correlation [

      Krabbe PFM. The measurement of health and health status: concepts, methods and applications from a multidisciplinary perspective [Internet], Version 1.0 [revised 2016 Oct 7; cited 2022 May 18]. Available from: https://doi.org/10.1016/B978-0-12-801504-9.00007-6.

      ] between consultative behavior and autonomous behavior (r = .24, p = .01) in the self-care management scale. This confirmed acceptable convergent validity for the two multidimensional scales.

      Internal consistency and item analysis

      Self-care maintenance scale

      When the internal consistency of the self-care maintenance was calculated with all 11 items using the 574-CFA sample, the alpha coefficient was .70 and McDonald's omega coefficient was .69. If the items were deleted, the omega coefficient ranged from .64 to .69; no item was expected to significantly increase the coefficient if deleted. When the multidimensional scale's composite reliability was tested [
      • Raykov T.
      Handbook of structural equation modeling. Reprint ed..
      ], the coefficient remained at .68, which is inadequate [
      • Nunnally J.
      • Bernstein I.H.
      Psychometric theory. 3rd rev. ed..
      ]. Most items presented adequate discrimination; the corrected item-total correlation of all items was >.30, excluding one item (scmt 04).

      Self-care management scale

      The self-care management scale's internal consistency was calculated with all five items using the 305-sample CFA data from participants who had experienced hypertension-related symptom within the last four weeks. The alpha coefficient was .62 and McDonald's omega coefficient was .65. The coefficient if items were deleted ranged from .49 to .68, while one item (scmt 01) was expected to significantly increase the coefficient if deleted. The composite reliability test on this multidimensional scale [
      • Raykov T.
      Handbook of structural equation modeling. Reprint ed..
      ] yielded a coefficient of .64, which was still inadequate [
      • Nunnally J.
      • Bernstein I.H.
      Psychometric theory. 3rd rev. ed..
      ]. Most items presented adequate discrimination, with the corrected item-total correlation of all items >.30, excluding one item (scmt 04).

      Self-care confidence scale

      The self-care maintenance scale's internal consistency was calculated with all six items using the 574-sample CFA dataset. The alpha coefficient was .90, and McDonald's omega coefficient was .89. The omega coefficient if items were deleted ranged from .87 to .88, with no item expected to significantly increase the coefficient if removed. All items presented adequate discrimination, with item to total corrected >.30.

      Discussion

      We evaluated the factorial validity, construct validity, and internal reliability of a Thai cross-cultural adaptation of the SC–HI (version 2.0). This is one of two recent studies outside the US to use EFA and CFA to evaluate the SC–HI. The dimensionality and psychometric properties we found partially supported the original model [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ]. The self-care maintenance scale had a multidimensional construct. Factor loadings on the self-care management scale's autonomous and consultative dimensions differed from those of previous studies [6.9]. The relationship between the Thai self-care confidence scale items differed from those of the Brazilian model [
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ]. The respecified self-care management and self-care confidence scales improved the goodness of fit indices, supporting construct validity. The HTMT ratios and Pearson correlation coefficients confirmed acceptable discriminant validity and convergent validity of a three-factor self-care maintenance, and a two-factor self-care management scale. We used a new method to estimate discriminant validity because traditional methods, such as the Fornall–Larcker criterion and cross-loading method, had unacceptably low sensitivity compared to the HTMT ratios [
      • Henseler J.
      • Ringle C.M.
      • Sarstedt M.
      A new criterion for assessing discriminant validity in variance-based structural equation modeling.
      ]. The self-care maintenance and self-care management scales' reliability coefficients were inadequate, whereas the self-care confidence reliability coefficient was acceptable. Overall, the study results illustrate that the cross-cultural adaptation of the Thai SC–HI is valid, but the reliability needs further testing.

      Self-care of hypertension inventory

      We used EFA to examine the dimensionality of the Thai SC–HI in this study, followed by CFA testing. We conducted EFA because it was already known that the scales characteristic of the US [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ] and international SC–HI versions [
      • Alsaqer K.
      • Bebis H.
      Cross-cultural adaptation, validity, and reliability of the Arabic version of the Self-care of Hypertension Inventory scale among older adults.
      ,
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ,
      • Zhao Q.
      • Guo Y.
      • Gu Y.
      • Yang L.
      Translation and cross-cultural adaptation of the Chinese version of the Self-care of Hypertension Inventory in older adults.
      ] differed. As expected, the Thai self-care maintenance and self-care management scales have a multidimensional structure, and the self-care confidence scale has a unidimensional structure. The multidimensional self-care maintenance scale in the Thai version does not support the previous models [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ,
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ]; however, the multidimensional self-care maintenance scale has psychometric characteristics in Chinese [
      • Zhao Q.
      • Guo Y.
      • Gu Y.
      • Yang L.
      Translation and cross-cultural adaptation of the Chinese version of the Self-care of Hypertension Inventory in older adults.
      ], Arabic [
      • Alsaqer K.
      • Bebis H.
      Cross-cultural adaptation, validity, and reliability of the Arabic version of the Self-care of Hypertension Inventory scale among older adults.
      ] and other relevant self-care measures [
      • Riegel B.
      • Barbaranelli C.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • Miller J.L.
      • et al.
      Development and initial testing of the self-care of chronic illness inventory.
      ,
      • Chen D.D.
      • Zhang H.
      • Cui N.
      • Tang L.
      • Shao J.
      • Wang X.
      • et al.
      Cross-cultural adaptation and validation of the caregiver contribution to self-care of chronic illness inventory in China: a cross-sectional study.
      ,
      • Vellone E.
      • Riegel B.
      • Cocchieri A.
      • Barbaranelli C.
      • D'Agostino F.
      • Glaser D.
      • et al.
      Validity and reliability of the caregiver contribution to self-care of heart failure index.
      ]. Although the original self-care management scale was a two-factor model [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ], items loaded on the autonomous and consultative dimensions in Thai model differ from those of the US and Brazilian models [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ,
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ]. The self-care confidence items covariance matrix differed for the Thai and Brazilian models. The modified Thai SC–HI model had a moderate to high factor loading matrix for all items in the final three scales; one item (scmn 02) from the US and Brazilian self-care management models had poor factor loading. Our findings confirmed the Thai SC–HI model's construct validity is consistent with other international SC–HI models [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ,
      • Alsaqer K.
      • Bebis H.
      Cross-cultural adaptation, validity, and reliability of the Arabic version of the Self-care of Hypertension Inventory scale among older adults.
      ,
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ,
      • Swiatoniowska-Lonc N.
      • Polanski J.
      • Jankowska-Polanska B.
      Psychometric properties of the Polish version of the self-care of hypertension inventory.
      ].
      The Self-care maintenance and self-care management reliability coefficients were slightly lower, but the self-care confidence reliability coefficient was adequate. We used McDonald's omega and a composite reliability coefficient for the multidimensional scales instead of Cronbach's alpha alone. For multidimensional scales, the Cronbach's alpha coefficient overestimates reliability for the general common factor and underestimates the reliability of all model factors [
      • Trizano-Hermosilla I.
      • Gálvez-Nieto J.L.
      • Alvarado J.M.
      • Saiz J.L.
      • Salvo-Garrido S.
      Reliability estimation in multidimensional scales: comparing the bias of six estimators in measures with a bifactor structure.
      ]. With Cronbach's alpha [
      • Swiatoniowska-Lonc N.
      • Polanski J.
      • Jankowska-Polanska B.
      Psychometric properties of the Polish version of the self-care of hypertension inventory.
      ], the highest coefficient values were observed in the self-care confidence scale followed by the self-care maintenance and self-care management scales. However, in the Chinese version [
      • Zhao Q.
      • Guo Y.
      • Gu Y.
      • Yang L.
      Translation and cross-cultural adaptation of the Chinese version of the Self-care of Hypertension Inventory in older adults.
      ], the lowest Cronbach's alpha coefficients were found in the self-care maintenance scale. The Cronbach's alpha differences may depend on individuals' sociodemographic backgrounds, and the manner in which they perform self-care. The self-care maintenance and self-care management scales' lower internal consistency could be expected because self-care behaviors are largely independent of each other; they are controlled by various motivators, personal and cultural aspects, and change over time [
      • Riegel B.
      • Barbaranelli C.
      • Carlson B.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • et al.
      Psychometric testing of the revised self-care of heart failure index.
      ].
      Self-care maintenance and self-care management items with lower corrected item-total correlation were expected to have high factor loading on other dimensions. The results were supported by both EFA and CFA. For example, we found that item scmt 01 “check your blood pressure” had the lowest score, because this self-care behavior is difficult to perform daily. Therefore, this finding supported deleting this item from the updated SC–HI version [
      • Dickson V.V.
      • Fletcher J.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory version 3.0.
      ]. Although the CFA and EFA samples had different characteristics, the model was considered valid. Also, this finding may demonstrate the Thai SC–HI's external validity in patients with various sociodemographic and clinical characteristics.

      Self-care maintenance

      Thai self-care maintenance was multidimensional, as theorized. CFA indicated better fit indices with either a three-factor or two-factor model, but not with a unidimensional model. The multidimensional structure is consistent with the Chinese SC–HI [
      • Zhao Q.
      • Guo Y.
      • Gu Y.
      • Yang L.
      Translation and cross-cultural adaptation of the Chinese version of the Self-care of Hypertension Inventory in older adults.
      ], the Self-Care of Chronic Illness Inventory (SC–CII) [
      • Riegel B.
      • Barbaranelli C.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • Miller J.L.
      • et al.
      Development and initial testing of the self-care of chronic illness inventory.
      ], and the Caregiver Contribution to Self-Care of Chronic Illness Inventory (CC–SC–CII) [
      • Chen D.D.
      • Zhang H.
      • Cui N.
      • Tang L.
      • Shao J.
      • Wang X.
      • et al.
      Cross-cultural adaptation and validation of the caregiver contribution to self-care of chronic illness inventory in China: a cross-sectional study.
      ,
      • Vellone E.
      • Riegel B.
      • Cocchieri A.
      • Barbaranelli C.
      • D'Agostino F.
      • Glaser D.
      • et al.
      Validity and reliability of the caregiver contribution to self-care of heart failure index.
      ], which were developed under similar theories [
      • Riegel B.
      • Jaarsma T.
      • Strömberg A.
      A middle-range theory of self-care of chronic illness.
      ]. We justified this scale's three-factor model based on KMO values and self-care actions' characteristics for each item and factor loading matrix. Items allocated to health promoting behaviors and illness related behaviors supported the self-care maintenance scale of the Self-Care of Chronic Illness Inventory [
      • Riegel B.
      • Barbaranelli C.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • Miller J.L.
      • et al.
      Development and initial testing of the self-care of chronic illness inventory.
      ]. Moreover, our findings revealed a third factor specific to medical care behaviors. The final three-factor model improved the fit indices and better explained the covariation effects in the two sets of three items. Each of the self-care behaviors correlated with the others. For example, patients trying to lose weight may eat a variety of vegetables, fruits, and grains (scmt02), and perform regular exercise (scmt 06).
      Our findings support the conceptual basis [
      • Riegel B.
      • Jaarsma T.
      • Strömberg A.
      A middle-range theory of self-care of chronic illness.
      ] of daily self-care maintenance to promote health, coupled with specific treatment to control hypertension. The Thai self-care maintenance dimensions supported the self-care of chronic illness theory and relevant self-care measurement in general [
      • Riegel B.
      • Jaarsma T.
      • Strömberg A.
      A middle-range theory of self-care of chronic illness.
      ,
      • Riegel B.
      • Barbaranelli C.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • Miller J.L.
      • et al.
      Development and initial testing of the self-care of chronic illness inventory.
      ,
      • Riegel B.
      • Barbaranelli C.
      • Carlson B.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • et al.
      Psychometric testing of the revised self-care of heart failure index.
      ]. Health promoting behaviors and illness related behaviors are two known dimensions [
      • Riegel B.
      • Barbaranelli C.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • Miller J.L.
      • et al.
      Development and initial testing of the self-care of chronic illness inventory.
      ,
      • Riegel B.
      • Barbaranelli C.
      • Carlson B.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • et al.
      Psychometric testing of the revised self-care of heart failure index.
      ]. Our study revealed another dimension, medical care behaviors. To control blood pressure and reduce cardiovascular risk, patients need to make lifestyle changes (e.g., diet, physical activity, weight control) and follow specific hypertension treatment protocols (e.g., medication, follow-up) [
      • Burnier M.
      • Egan B.M.
      Adherence in hypertension.
      ,
      • Cornelissen V.A.
      • Smart N.A.
      Exercise training for blood pressure: a systematic review and meta-analysis.
      ,
      • Cook N.R.
      • Cutler J.A.
      • Obarzanek E.
      • Buring J.E.
      • Rexrode K.M.
      • Kumanyika S.K.
      • et al.
      Long term effects of dietary sodium reduction on cardiovascular disease outcomes: observational follow-up of the trials of hypertension prevention (TOHP).
      ].

      Self-care management

      Self-care management was bidimensional based on theory and in the original SC–HI [
      • Riegel B.
      • Jaarsma T.
      • Strömberg A.
      A middle-range theory of self-care of chronic illness.
      ,
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ]. Our analysis revealed autonomous and consultative dimension item numbers and factor loadings that differed from those of the US [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ] and Brazilian [
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ] models. The factorial structure supported the Brazilian model and reflected cultural diversity. Autonomous dimension items focused on controlling symptoms, while consultative dimension items focused on daily self-care promoting behaviors. The controlling symptoms dimension in this study was similar to the self-care of heart failure index, which adequately addresses the theory of self-care, but it is not a self-care management subscale [
      • Riegel B.
      • Barbaranelli C.
      • Carlson B.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • et al.
      Psychometric testing of the revised self-care of heart failure index.
      ]. This finding supported a new revised SC–HI component [
      • Dickson V.V.
      • Fletcher J.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory version 3.0.
      ] that separates items relevant to the self-care monitoring scale from the alternative self-care management scale.
      Interestingly, CFA confirmed a good fit and supported all Thai, US, and Brazilian models. Our model had a better factor loading matrix with all items, specifically recognizing symptoms (item: scmn 12). The Self-Care Management process could be dissimilar but share commonality in one experience. Response to body changes may be related to emotion and illness. However, the response to change depends on physiological or disease/health related personal, psychological, sociological, and developmental factors [
      • Riegel B.
      • Jaarsma T.
      • Lee C.S.
      • Strömberg A.
      Integrating symptoms into the middle-range theory of self-care of chronic illness.
      ]. Cross-cultural adaptations of the SC–HI must verify self-care management structures for the targeted population. Unlike other cardiovascular diseases, such as heart failure [
      • Riegel B.
      • Barbaranelli C.
      • Carlson B.
      • Sethares K.A.
      • Daus M.
      • Moser D.K.
      • et al.
      Psychometric testing of the revised self-care of heart failure index.
      ], hypertension has less severe symptoms. Common hypertension symptoms include headache, dizziness, and visual impairment; however, these symptoms are not disease specific and may occur in normotensive individuals. Therefore, hypertensive patients may not know when their blood pressure is elevated because they might not feel this change. Timely response to uncontrolled blood pressure and adequate behavioral change may depend on patient awareness, interpretation, and recognition of hypertensive conditions [
      • Riegel B.
      • Jaarsma T.
      • Lee C.S.
      • Strömberg A.
      Integrating symptoms into the middle-range theory of self-care of chronic illness.
      ].

      Self-care confidence

      We found that self-care confidence was a unidimensional construct similar to the original model [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ]. Despite the high factor loading matrix of all six items, the model partially supports most fit indices (CFI, TLI, and SRMR), except RMSEA. Unlike the Brazilian model [
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ], we found a high covariance between items scc 18 and scc 19 and scc 18 and scc 20. The final modified model improved all relevant fit indices and better explained the covariation effects among these items. This correlation was not present in the US [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ], Chinese [
      • Zhao Q.
      • Guo Y.
      • Gu Y.
      • Yang L.
      Translation and cross-cultural adaptation of the Chinese version of the Self-care of Hypertension Inventory in older adults.
      ], or Polish [
      • Swiatoniowska-Lonc N.
      • Polanski J.
      • Jankowska-Polanska B.
      Psychometric properties of the Polish version of the self-care of hypertension inventory.
      ] models; however, the Brazilian model [
      • Silveira L.C.J.
      • De Maria M.
      • Dickson V.V.
      • Avila C.W.
      • Rabelo-Silva E.R.
      • Vellone E.
      Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
      ] showed correlation between items scc 22 “evaluate changes in your blood pressure” and scc 23 “take action that will control your blood pressure.” Since the self-care confidence items scc 18, scc 19, and scc 20 measured different actions, our model demonstrated the relationship between confidence in blood pressure control, following a treatment regimen, and recognizing changes in health. Self-care confidence is a motivating factor for self-care maintenance and self-care management. The scale is a promising measure of self-efficacy within a health belief model; individuals' confidence facilitates specific self-care tasks [
      • Dickson V.V.
      • Lee C.
      • Yehle K.S.
      • Abel W.M.
      • Riegel B.
      Psychometric testing of the self-care of hypertension inventory.
      ]. Higher self-efficacy is associated with better hypertension self-care behaviors, including dietary changes, physical activity, and medical adherence [
      • Tan F.C.J.H.
      • Oka P.
      • Dambha-Miller H.
      • Tan N.C.
      The association between self-efficacy and self-care in essential hypertension: a systematic review.
      ].

      Limitation and strengths

      One limitation is that the self-care maintenance and self-care management scales' reliability coefficients were only partially accepted. This might be because factor loadings for the two items were lower than desired (items: scmt 01 “check your blood pressure,” scmn 15 “be strict on taking your blood pressure-lowering medicines”). However, because the indexes reliability in the population is unknown, we cannot know whether these conditions were met, hence further testing is needed to identify areas with compromised reliability in a national-level hypertension population. Also, we were unable to provide evidence for concurrent validity in this analysis because we lacked adequate data and an appropriate instrument. A third limitation is that the majority of the participants were recruited from rural primary care facilities, where lifestyle, behaviors, and environment may differ from those of urban or inner-city area. Study strengths include: participant enrollment from several settings; large sample sizes that were adequate for psychometric property testing; and participants (females, older adults, socioeconomic advantages, and various cardiometabolic risk factors) who were generally representative of Thailand's hypertensive population. We used two data sets: one to explore factorial validity using EFA, and the second to confirm construct validity. Thus, our analysis may reflect both hypothesis testing and generalizability. Implications for future research include testing those who may require hypertension self-care for complicated health conditions, such as older adults. Concurrent validity is also needed for full scale evaluation. Further research is needed to verify that the measure predicts hypertension results internationally.

      Conclusion

      The Thai SC–HI is a valid reflection of the theoretical concept on which it is based. This instrument is a useful clinical tool to assess and guide self-care development for promoting optimal health, hypertension treatment regimens, symptom management, and self-care confidence for Thai individuals with hypertension.

      Ethical consideration

      Approval was obtained from the Walailak Ethics Board Committee (code number: 59/075). The study adheres to standards delineated in the Declaration of Helsinki.

      Funding

      The research was financially supported by Walailak University under the National Research Council of Thailand (NRCT grant #WU60115 ), the Excellent Center of Community Health Promotion of Walailak University , and the new strategies research project (P2P) fiscal year 2022, Walailak University, Thailand.

      Data statement

      The original study is registered at Open Science Framework (https://osf.io/8j95k/).

      Conflict of interest

      The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

      Acknowledgment

      We thank Professor Barbara Riegel, PhD, RN, FAHA, FHFSA, FAAN, for reviewing and approving Thai SC-HI, and also for encouraging us to conduct the psychometric test.

      References

        • Aekplakorn W.
        • Hathaichanok P.
        • Kanittha T.
        • Warapone S.
        The 5th national health examination survey.
        Health System Research Institute, Nonthaburi20142016: 2557 ([cited 2022 Apr 2]. Available from:)
        • Burnier M.
        • Egan B.M.
        Adherence in hypertension.
        Circ Res. 2019; 124: 1124-1140
        • Cornelissen V.A.
        • Smart N.A.
        Exercise training for blood pressure: a systematic review and meta-analysis.
        J Am Heart Assoc. 2013; 2e004473
        • Cook N.R.
        • Cutler J.A.
        • Obarzanek E.
        • Buring J.E.
        • Rexrode K.M.
        • Kumanyika S.K.
        • et al.
        Long term effects of dietary sodium reduction on cardiovascular disease outcomes: observational follow-up of the trials of hypertension prevention (TOHP).
        BMJ. 2007; 334: 885-888
        • Riegel B.
        • Jaarsma T.
        • Strömberg A.
        A middle-range theory of self-care of chronic illness.
        Adv Nurs Sci. 2012; 35: 194-204
        • Dickson V.V.
        • Lee C.
        • Yehle K.S.
        • Abel W.M.
        • Riegel B.
        Psychometric testing of the self-care of hypertension inventory.
        J Cardiovasc Nurs. 2017; 32: 431-438
        • Phonphet C.
        • Suwanno J.
        • Thiamwong L.
        • Mayurapak C.
        • Ninla-Aesong P.
        Translation and cross-cultural adaptation of the self-care of hypertension inventory for Thais with hypertension.
        J Cardiovasc Nurs. 2022; (In press)
        • Alsaqer K.
        • Bebis H.
        Cross-cultural adaptation, validity, and reliability of the Arabic version of the Self-care of Hypertension Inventory scale among older adults.
        J Cardiovasc Nurs. 2021; 36: 430-436
        • Silveira L.C.J.
        • De Maria M.
        • Dickson V.V.
        • Avila C.W.
        • Rabelo-Silva E.R.
        • Vellone E.
        Validity and reliability of the self-care of hypertension inventory (SC-HI) in a Brazilian population.
        Heart Lung. 2020; 49: 518-523
        • Zhao Q.
        • Guo Y.
        • Gu Y.
        • Yang L.
        Translation and cross-cultural adaptation of the Chinese version of the Self-care of Hypertension Inventory in older adults.
        J Cardiovasc Nurs. 2019; 34: 124-129
        • Ma Y.
        • Cheng H.Y.
        • Sit J.W.H.
        • Chien W.T.
        Psychometric evaluation of the Chinese version of hypertension self-care profile.
        J Cardiovasc Nurs. 2021; 36: 420-429https://doi.org/10.1097/JCN.0000000000000708
        • Wee S.Y.
        • Salim H.
        • Mawardi M.
        • Koh Y.L.E.
        • Ali H.
        • Shariff Ghazali S.
        • et al.
        Comparing and determining factors associated with hypertension self-care profiles of patients in two multi-ethnic Asian countries: cross-sectional studies between two study populations.
        BMJ Open. 2021; 11e044192https://doi.org/10.1136/bmjopen-2020-044192
        • Lee J.E.
        • Han H.R.
        • Song H.
        • Kim J.
        • Kim K.B.
        • Ryu J.P.
        • et al.
        Correlates of self-care behaviors for managing hypertension among Korean Americans: a questionnaire survey.
        Int J Nurs Stud. 2010; 47: 411-417https://doi.org/10.1016/j.ijnurstu.2009.09.011
        • Riegel B.
        • Barbaranelli C.
        • Sethares K.A.
        • Daus M.
        • Moser D.K.
        • Miller J.L.
        • et al.
        Development and initial testing of the self-care of chronic illness inventory.
        J Adv Nurs. 2018; 74: 2465-2476https://doi.org/10.1111/jan.13775
        • Osborne J.W.
        Best practice in exploratory factor analysis.
        CreateSpace Independent Publishing, California, Scotts Valley2014
        • Raykov T.
        Handbook of structural equation modeling. Reprint ed..
        The Guilford Press, New York, NY2012: 472-492 (Chapter 28, Scale construction and development using structural equation modeling)
        • McDonald R.P.
        Test theory: a unified treatment.
        1st ed. Lawrence Erlbaum Associates, New Jersey, Mahwah1999
        • Trizano-Hermosilla I.
        • Gálvez-Nieto J.L.
        • Alvarado J.M.
        • Saiz J.L.
        • Salvo-Garrido S.
        Reliability estimation in multidimensional scales: comparing the bias of six estimators in measures with a bifactor structure.
        Front Psychol. 2021; 12508287
        • Barbaranelli C.
        • Lee C.S.
        • Vellone E.
        • Riegel B.
        Dimensionality and reliability of the self-care of heart failure index scales: further evidence from confirmatory factor analysis.
        Res Nurs Health. 2014; 37: 524-537
        • Riegel B.
        • Dickson V.V.
        • Faulkner K.M.
        The situation-specific theory of heart failure self-care: revised and updated.
        J Cardiovasc Nurs. 2016; 31: 226-235
        • Anthoine E.
        • Moret L.
        • Regnault A.
        • Sebille V.
        • Hardouin J.B.
        Sample size used to validate a scale: a review of publications on newly-developed patient reported outcomes measures.
        Health Qual Life Outcomes. 2014; 12: 176https://doi.org/10.1186/s12955-014-0176-2
        • Tabachnick B.G.
        • Fidell L.S.
        Using multivariate statistics.
        5th ed. Pearson Education, Massachusetts, Boston2007
        • Muthén B.
        • Kaplan D.A.
        Comparison of some methodologies for the factor analysis of non-normal Likert variables.
        Br J Math Stat Psychol. 1985; 38: 171-189https://doi.org/10.1111/j.2044-8317.1985.tb00832.x
        • Kaiser H.F.
        An index of factorial simplicity.
        Psychometrika. 1974; 39: 31-36https://doi.org/10.1007/BF02291575
        • Riegel B.
        • Barbaranelli C.
        • Carlson B.
        • Sethares K.A.
        • Daus M.
        • Moser D.K.
        • et al.
        Psychometric testing of the revised self-care of heart failure index.
        J Cardiovasc Nurs. 2019; 34: 183-192
        • Hu L.
        • Bentler P.M.
        Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives.
        Struct Equat Model. 1999; 6: 1-55
        • Browne M.W.
        • Cudek R.
        Alternative ways of assessing model fit.
        Sociol Methods Res. 1992; 21: 230-258https://doi.org/10.1177/0049124192021002005
      1. Krabbe PFM. The measurement of health and health status: concepts, methods and applications from a multidisciplinary perspective [Internet], Version 1.0 [revised 2016 Oct 7; cited 2022 May 18]. Available from: https://doi.org/10.1016/B978-0-12-801504-9.00007-6.

        • Hubley A.M.
        Discriminant validity.
        in: Michalos A.C. Encyclopedia of quality of life and well-being research [Internet]. Springer, Dordrecht2014https://doi.org/10.1007/978-94-007-0753-5_751 ([cited 2002 May 18]. Available from:)
        • Henseler J.
        • Ringle C.M.
        • Sarstedt M.
        A new criterion for assessing discriminant validity in variance-based structural equation modeling.
        J Acad Mark Sci. 2015; 43: 115-135https://doi.org/10.1007/s11747-014-0403-8
        • Ronkko M.
        • Cho E.
        An updated guideline for assessing discriminant validity.
        Organ Res Methods. 2020; 25: 6-47
        • Bagozzi R.P.
        • Yi Y.
        Specification, evaluation, and interpretation of structural equation models.
        J Acad Mark Sci. 2012; 40: 8-34
        • Nunnally J.
        • Bernstein I.H.
        Psychometric theory. 3rd rev. ed..
        McGraw Hill, Inc, New York, NY1994
        • Chen D.D.
        • Zhang H.
        • Cui N.
        • Tang L.
        • Shao J.
        • Wang X.
        • et al.
        Cross-cultural adaptation and validation of the caregiver contribution to self-care of chronic illness inventory in China: a cross-sectional study.
        BMJ Open. 2021; 11e048875
        • Vellone E.
        • Riegel B.
        • Cocchieri A.
        • Barbaranelli C.
        • D'Agostino F.
        • Glaser D.
        • et al.
        Validity and reliability of the caregiver contribution to self-care of heart failure index.
        J Cardiovasc Nurs. 2013; 28: 245-255
        • Swiatoniowska-Lonc N.
        • Polanski J.
        • Jankowska-Polanska B.
        Psychometric properties of the Polish version of the self-care of hypertension inventory.
        J Cardiovasc Nurs. 2021; 36: 437-445
        • Dickson V.V.
        • Fletcher J.
        • Riegel B.
        Psychometric testing of the self-care of hypertension inventory version 3.0.
        J Cardiovasc Nurs. 2021; 36: 411-419
        • Riegel B.
        • Jaarsma T.
        • Lee C.S.
        • Strömberg A.
        Integrating symptoms into the middle-range theory of self-care of chronic illness.
        Adv Nurs Sci. 2019; 42: 206-215
        • Tan F.C.J.H.
        • Oka P.
        • Dambha-Miller H.
        • Tan N.C.
        The association between self-efficacy and self-care in essential hypertension: a systematic review.
        BMC Fam Pract. 2021; 22: 44