ORIGINAL RESEARCH

Structural equation model testing the situation-specific theory of

heart failure self-care

Ercole Vellone, Barbara Riegel, Fabio D’Agostino, Roberta Fida, Gennaro Rocco,

Antonello Cocchieri & Rosaria Alvaro

Accepted for publication 9 February 2013

Correspondence to E. Vellone:

e-mail: [email protected]

Ercole Vellone MSN RN

Research fellow

School of Nursing, University Tor Vergata,

Rome, Italy

Barbara Riegel DNSc RN FAAN

Professor

School of Nursing, University of

Pennsylvania, Philadelphia, USA

Fabio D’Agostino MSN RN

PhD candidate

School of Nursing, University Tor Vergata,

Rome, Italy

Roberta Fida PhD

Assistant Professor

Department of Psychology, “Sapienza”

University, Rome, Italy

Gennaro Rocco MSN RN

President

Center of Excellence for Nursing

Scholarship, Rome, Italy

Antonello Cocchieri MSN RN

PhD candidate

School of Nursing, University Tor Vergata,

Rome, Italy

Rosaria Alvaro MSN RN

Associate Professor

School of Nursing, University Tor Vergata,

Rome, Italy

VELLONE E., R I EGEL B . , D ’AGOST INO F . , F IDA R . , ROCCO G . , COCCH IER I A .

& ALVARO R . ( 2 0 1 3 ) Structural equation model testing the situation-specific

theory of heart failure self-care. Journal of Advanced Nursing 69(11), 2481–

2492. doi: 10.1111/jan.12126

Abstract Aim. To test the situation-specific theory of heart failure self-care with structural

equation modelling.

Background. Several authors have proposed theories on heart failure self-care,

but only the situation-specific theory of heart failure self-care by Riegel and

Dickson is focused on the process that patients use to perform self-care. This

theory has never been tested with structural equation modelling.

Design. A secondary analysis of data from a cross-sectional study.

Methods. Patients with heart failure were recruited in 21 cardiovascular centres

across Italy during 2011. Data were collected with a sociodemographic

questionnaire, chart abstraction for clinical data and the Self-Care of Heart

Failure Index v.6�2. Results. A sample of 417 participants was enrolled in the study (59% males,

mean age 72 years). The following propositions were tested and supported:

Symptom monitoring correlates with treatment adherence; symptom monitoring

and treatment adherence have a direct, positive relationship with symptom

recognition and evaluation that in turn have a direct, positive relationship with

treatment implementation; treatment implementation has a direct, positive

relationship with treatment evaluation. In addition, the following three

relationships were found: Symptom monitoring has a direct, positive relationship

with treatment implementation; symptom recognition and evaluation have direct,

positive relationships with treatment evaluation and symptom monitoring

correlates with treatment evaluation. [Correction added on 9th April 2013, after

first online publication: ‘. . .symptom monitoring correlates with treatment

implementation.’ has been corrected to read ‘. . .symptom monitoring correlates

with treatment evaluation.’]

Conclusion. The data support the situation-specific theory of heart failure self-

care with the addition of three new relationships that emerged from the analysis.

Results of this study lend further support to the use of the situation-specific

theory of heart failure self-care in research and practice.

© 2013 Blackwell Publishing Ltd 2481

JAN JOURNAL OF ADVANCED NURSING

Keywords: heart failure, nursing, self-care, structural equation modelling, symp-

tom monitoring, symptom recognition and evaluation, theory testing, treatment

adherence, treatment implementation

Introduction

Heart Failure (HF) is the most common cardiovascular dis-

ease in many countries worldwide (Caldarola et al. 2009,

Jiang & Ge 2009, Ntusi & Mayosi 2009, Norton et al.

2011). It is estimated that 6�6 million North Americans

(Roger et al. 2012) and 15 million Europeans (Anguita

Sanchez et al. 2008) are affected by HF. The prevalence of

HF is constantly increasing due to the ageing of the popula-

tion, improved treatment, and survival rates after myocar-

dial infarction and the continuing problem of poor control

of hypertension.

Heart failure patients experience lower quality of life

than patients affected by other chronic conditions (Juenger

et al. 2002, Iavazzo & Cocchia 2011, Burstrom et al.

2012) and are prone to frequent hospitalization and emer-

gency department visits for illness decompensation (Krum-

holz et al. 2009, Ross et al. 2010). Mortality remains high

with about the 30% of people with HF dying within the

first year after diagnosis (Barsheshet et al. 2010, Chen et al.

2011).

Self-care of HF is considered essential to improving

patients’ quality of life and reducing hospitalization,

mortality, and emergency department visits (Bird et al.

2010, Buck et al. 2012). In the last two decades several

authors have proposed theories of self-care for use in

research and clinical practice. While all these theories

identify the components and predictors of HF self-care,

only the situation-specific theory by Riegel and Dickson

(2008) has specifically focused on the process that HF

patients use in the performance of self-care (Figure 1).

Although this theory is widely cited no study testing the

relationships among the theoretical concepts was

located.

Background

Theories of self-care in heart failure

Meleis (2011) defines theory as a coherent vision of the

context, process, and outcomes associated with a specific

phenomenon. As demonstrated below, numerous nursing

investigators have proposed models of HF self-care with

variable attention given to these elements of theory.

In studying self-care behaviours of people with HF,

Jaarsma et al. (2000), used three sets of self-care limitations

from Orem’s theory of self-care: knowledge, judging and

decision making, and action and result achievement. Later,

Orem’s theory was used by Jaarsma et al. (2003) to develop

the European Heart Failure Self-care Behaviour Scale

(EHFScBS). In this effort, HF self-care was specified as

involving three constructs: complying with the regimen,

(e.g. daily weighing, sodium and fluid restriction), asking

for help (e.g. call the doctor/nurse in case of weight gain or

excessive fatigue), and adapting activities (e.g. resting).

These three constructs, although describing the components

of self-care, do not represent a theory of HF self-care where

concepts are linked with propositions to explain a process.

Granger et al. (2006) used the middle-range Trajectory of

Chronic Illness Theory (TCIT) by Strauss et al. (1984) to

integrate patients’ perspectives in self-care with those of HF

providers. The TCIT evolved from ethnographic work with

patients affected by chronic illnesses. This theory conceptu-

alizes relationships among factors contributing to the man-

agement of illness and the target therapeutic interventions.

According to this theory patients have their own perception

of the illness; they interpret and report symptoms and per-

ceive prescribed medications differently from healthcare

professionals. Using the TCIC, clinicians can integrate their

perspectives with those of patients. The principal concepts

Symptom monitoring

Symptom recognition

and evaluation

Treatment implementation

Treatment evaluation

Treatment adherence

Figure 1 The situation-specific theory of

heart failure self-care showing the rela-

tionship between Self-care Maintenance

and Self-care Management.

2482 © 2013 Blackwell Publishing Ltd

E. Vellone et al.

13652648, 2013, 11, D ow

nloaded from https://onlinelibrary.w

iley.com /doi/10.1111/jan.12126 by R

egis C ollege, W

iley O nline L

ibrary on [25/03/2023]. See the T erm

s and C onditions (https://onlinelibrary.w

iley.com /term

s-and-conditions) on W iley O

nline L ibrary for rules of use; O

A articles are governed by the applicable C

reative C om

m ons L

icense

of the TCIC include the trajectory, that is the illness course,

the trajectory projection reflecting the goals of care, the tra-

jectory schema or the regimen for reaching the goals of

care, the trajectory management specifying how the regimen

is carried out, the conditions influencing management that

are the personal, interpersonal, and social contexts that

influence the regimen and the trajectory phasing, which is

the ups and downs of the clinical outcomes. Although this

theory can be considered a valuable tool to understand the

illness trajectory, it is not specific to HF self-care.

Bennett et al. (2001) developed the Beliefs about Medica-

tion Compliance Scale (BMCS) and the Beliefs about Die-

tary Compliance Scale (BDCS) for patients with HF. Both

instruments were based on the Health Belief Model (HBM)

that attempts to explain and predict health behaviours using

a focus on the individual’s attitudes and beliefs. From the

HBM, these authors took only the concepts of perceived

benefits and barriers, which were applied only to percep-

tions about water pills and the low-salt diet and not to other

self-care activities. In addition, the authors did not elaborate

a mechanism to explain how self-care works in HF.

From the HBM, Connelly (Connelly 1987, 1993) devel-

oped the Model of Self-Care in Chronic Illness (MSCCI)

that was modified and tested in people with HF (Rockwell

& Riegel 2001). The authors of this study conceptualized

that general and therapeutic self-care behaviours are influ-

enced by predisposing variables (self-concept, health moti-

vations, and patient perceptions) and enabling variables

(patient characteristics, psychological status, regimen fea-

tures, cue to action, social support, and system characteris-

tics). Study results showed that only educational level and

the severity of symptoms explained HF self-care. Although

the concepts of self-care maintenance and self-care manage-

ment were described in this article and the investigators

identified variables influencing self-care, they did not

explain the process of self-care per se or how self-care

maintenance related to self-care management.

Moser and Watkins (2008) described five factors affect-

ing decision making and subsequently self-care maintenance

and self-care management in HF patients in a life course

model. The five factors were health literacy, psychological

status, symptom status, ageing status, prior experiences

with symptoms, and the healthcare system. Although this

work gave an important overview of the factors affecting

decision-making and self-care in HF, the manner where the

variables relate to each other was not considered.

In early work, Riegel et al. (2000) described a process of

self-management of HF that later developed into the situa-

tion-specific theory of HF self-care (Riegel & Dickson 2008).

According to the situation-specific theory, self-care is a

naturalistic decision-making process that includes self-care

maintenance and self-care management (Figure 1). Self-care

maintenance refers to symptom monitoring (checking weight

and ankle for swelling) and treatment adherence (e.g. low salt

diet, keeping health provider appointment, exercising) that

reflect behaviours used to maintain physiological stability.

Self-care maintenance, considered the base of self-care, influ-

ences self-care management. Self-care management is a com-

plex process that requires HF patients to act when symptoms

of exacerbation occur, particularly ankle swelling and breath-

ing problems. Self-care management has been described as

being composed of symptom recognition, symptom evalua-

tion, treatment implementation and treatment evaluation.

These actions have been theorized as occurring in sequence,

so symptom recognition influences treatment implementation

and treatment implementation influences treatment evalua-

tion. According to Riegel, the self-care process is influenced

by confidence in one’s ability to perform self-care. As the situ-

ation-specific theory of HF self-care is most highly developed

and an instrument exists with which to measure the various

components of the process, we used structural equation mod-

elling (SEM) to improve our understanding of the process of

HF self-care and of the relationships among the theoretical

concepts.

The study

Aim

The aim of this study was to test the situation-specific theory

of HF self-care with SEM. Such testing would improve

knowledge of the process of HF self-care and of the relation-

ships among the theoretical concepts of treatment adherence,

symptom monitoring, symptom recognition and evaluation,

treatment implementation and treatment evaluation.

Research hypothesis

The overarching hypothesis was that the model would fit

the data, but the following specific hypotheses derived from

the situation-specific theory of HF self-care (Figure 1) were

tested as well:

● Symptom monitoring correlates with treatment adher-

ence.

● Symptom monitoring and treatment adherence have

direct, positive relationships with symptom recognition

and evaluation.

● Symptom recognition and evaluation have direct, posi-

tive relationships with treatment implementation.

© 2013 Blackwell Publishing Ltd 2483

JAN: ORIGINAL RESEARCH Testing the situation-specific theory of HF self-care

13652648, 2013, 11, D ow

nloaded from https://onlinelibrary.w

iley.com /doi/10.1111/jan.12126 by R

egis C ollege, W

iley O nline L

ibrary on [25/03/2023]. See the T erm

s and C onditions (https://onlinelibrary.w

iley.com /term

s-and-conditions) on W iley O

nline L ibrary for rules of use; O

A articles are governed by the applicable C

reative C om

m ons L

icense

● Treatment implementation has a direct, positive rela-

tionship with treatment evaluation.

Design

A secondary analysis of data from a cross sectional study

was used.

Participants

A convenience sample of 659 participants was enrolled.

From these 659, 417 patients with data on self-care mainte-

nance and self-care management were included. Subjects

excluded at this point were typically missing data on self-

care management because this scale can be measured only

in symptomatic patients. All participants were at least

18 years of age and had a confirmed diagnosis of HF. The

confirmed diagnosis of HF was established using the diag-

nostic criteria specified by the European Society of Cardiol-

ogy guidelines (Dickstein 2008), reconfirmed in 2012

(McMurray et al. 2012). In addition, patients without a

coronary event in the last three months were selected based

on the rationale that soon after a coronary event patients

might find it difficult to perform physical exercise (a com-

ponent of self-care). Patients were recruited from 21 cardio-

vascular ambulatory clinics or day hospitals across Italy.

Data collection

Instruments

The following instruments were used to collect the data.

The sociodemographic questionnaire. This survey was

designed by the research team, even though most items

have been used repeatedly in other studies (Riegel et al.

2010a, Vellone et al. 2012b) to collect age, gender, mari-

tal status, job, educational level, New York Heart Associ-

ation (NYHA) class, ejection fraction, and time since

diagnosis. Functional class measured with the New York

Heart Association (NYHA) scale, ejection fraction, and

time since diagnosis were abstracted from the patient’s

clinical record.

The Self-care of Heart Failure Index version 6�2 (SCHFI

v.6�2) (Riegel et al. 2009). It is a widely used measure of

HF self-care. The instrument is composed of three scales:

(i) the self-care maintenance scale (ten items) measures

symptom monitoring (two items), and treatment adherence

(eight items); (ii) the self-care management scale (six items)

measures HF patients’ actions and responses when symp-

toms occur and specifically symptom recognition and evalu-

ation (one item), treatment implementation (four items),

and treatment evaluation (one item); (iii) the self-care confi-

dence scale (six items) evaluates confidence in each of the

self-care processes, but this scale was not used in the analy-

sis since self-care confidence is not a component of self-care

but instead a factor that influences self-care (Riegel et al.

2009). The 22 item SCHFI v.6�2 uses a 4-point self-report

scale from Never or Rarely to Always or Daily. Three sepa-

rate scores can be computed from this index, all of which

have a possible range of 0–100, the higher the score the

better the self-care.

For the purposes of this study, the individual items were

aggregated conceptually as shown in Table 1 to obtain con-

ceptual measures that could be used to model the theoreti-

cal structure of the situation-specific theory. Each of these

Table 1 Conceptual aggregations of the SCHFI v.6�2 items

Conceptual components Definitions SCHFI v.6�2 item contents

Symptom monitoring Actions patients engage in to monitor HF symptoms and to

prevent HF exacerbation

Daily weighing

Ankle checking for swelling

Treatment adherence Actions patients engage in to follow the HF treatment plan

and to live a healthy life

Following low-salt diet

Taking medication as prescribed

Attending health care provider

Doing physical activities

Using systems to remind to take medicine

Symptom recognition and

evaluation

Recognition and evaluation of changes in health status

related to HF

Time for recognition ankle swelling and problem

breathing as HF symptoms

Treatment implementation Decision to take action and implement treatments in

case of HF symptoms

Likelihood patients do the following actions in

case of ankle swelling or problem breathing:

-reducing salt in diet;

-drink less water;

-taking an extra diuretic;

-calling healthcare provider to ask for advice.

Treatment evaluation Evaluation of the actions taken to treat HF symptoms Being sure that implemented treatment helped

of not helped the patient

2484 © 2013 Blackwell Publishing Ltd

E. Vellone et al.

13652648, 2013, 11, D ow

nloaded from https://onlinelibrary.w

iley.com /doi/10.1111/jan.12126 by R

egis C ollege, W

iley O nline L

ibrary on [25/03/2023]. See the T erm

s and C onditions (https://onlinelibrary.w

iley.com /term

s-and-conditions) on W iley O

nline L ibrary for rules of use; O

A articles are governed by the applicable C

reative C om

m ons L

icense

conceptual aggregates was standardized to a 0–100 point

scale to be consistent with scoring of the original measure.

Procedure of data collection

Participants signed informed consent after the study was

explained by research assistants, all of whom were regis-

tered nurses. Sometimes patients completed the instruments

on their own but often the research assistants assisted in

instrument completion. The same research assistants

abstracted clinical records to obtain information, such as

NYHA class, ejection fraction, and time since diagnosis.

Data were collected during 2011.

Ethical consideration

The Institutional Review Board of each centre approved the

study before data collection.

Data analysis

Descriptive statistics were used to describe the sociodemo-

graphic and clinical characteristics of the sample (mean, SD,

ranges, median, and interquartile ranges) and were used to

analyse the component scores. The relationships among the

theory components were analysed by Pearson’s r. Then the

hypothesized model (Figure 1) was tested using SEM. We used

SEM because it is particularly well-suited for simultaneously

investigating the nomological network among the different

constructs specified in the model. In this network, the first

series of paths corresponds to the posited relationship

between symptom monitoring and treatment adherence as

independent variables and symptom recognition and evalua-

tion as the dependent variable. A second series of paths corre-

sponds to the posited relationship between symptom

recognition and evaluation as an independent variable and

treatment implementation as the dependent variable. Finally,

a third series of paths corresponds to the posited relationship

between treatment implementation as independent variable

and treatment evaluation as the dependent variable

(Figure 1). This statistically powerful approach allowed us to

investigate the mediating role of symptom recognition and

evaluation and treatment implementation, which simulta-

neously act as both dependent and independent variables.

Using a multifaceted approach to the assessment of the

model fit (Tanaka 1993), taking into account the recommen-

dations of Hu and Bentler (Hu & Bentler 1998, 1999), the

following fit indices were considered: (i) chi square, (ii) Com-

parative Fit Index (CFI; (Bentler 1990)), (iii) Root Mean

Square Error of Approximation (RMSEA; (Steiger 1990)),

and (iv) Standardized Root Mean Square Residual (SRMR;

(J€oreskog & S€orbom 1993)). Overall model fit was judged

using these cut-off values: CFI � 0�95 (Hu & Bentler 1999),

RMSEA up to 0�05 and in the lower bound of the 90% CI

(Browne & Cudek 1993) and SRMR values below 0�08 (Hu & Bentler 1998, 1999) as indicating a good fit.

Power analyses for SEM models are complicated and often

rest on assumptions that are impractical or not viable. We

followed the practice recommended by Jaccard and Turrisi

(Jaccard &Wan 1996) that provides a rough sense of statisti-

cal power by applying power analytic methods for ordinary

least squares regression as applied to selected linear equations

from the set of linear equations implied by the model in ques-

tion. To determine an appropriate sample size, in fact, struc-

tural equation modeling requires that in addition to

statistical power, issues of the stability of the covariance

matrix and the use of asymptotic theory be taken into

account. In terms of power, it is difficult to evaluate the

power associated with specific path coefficients in complex

SEM models because of the large number of assumptions

about population parameters that must be made. A rough

approximation of power can be obtained by using a limited

information approach with single indicators of the path mod-

els implied by Figure 1. This permits the use of traditional

power analysis software to gain a sense of sample size

demands (Jaccard & Wan 1996). For a multiple regression

analysis with four predictors where the squared multiple cor-

relation is 0�30 and where one wants to detect a predictor

that accounts for at least 5% unique variance in the outcome,

the required sample size to achieve power of 0�80 is approxi-

mately 115. Moreover, Barret (Barret 2007) suggested the

use of the rule of a minimum of 200 subjects, since power

analysis is too complex in SEM. Overall our sample size of

<p