Association between Self-Efficacy and Health Behaviour in Disease Control : A Systematic Review

Background: Self-efficacy is defined as his or her belief of their capacity to produce specific performance attainments where represent the ability of positive and negative control over their own motivation, behavior, and social environment. Interventions to increase self-efficacy are a potentially effective way of changing health behavior towards attractive results, however the subject has not been systematically reviewed. This review aims to assess the relationship between self-efficacy and change in good health behaviour. Methods: A comprehensive and extensive search of four bibliographic data bases was conducted for papers reporting health promotion and educational interventions that explicitly targeted self-efficacy in order to change health behaviour in achieving better disease control. Results: Out of 314 studies, 13 were included in the review. Of these, 8 were found to have significant association between self-efficacy and specific health behaviours. However, 5 studies failed to show any significant prediction towards health behavior. Another 6 studies showed either significant mediation effect or indirect relationship of self-efficacy and health behaviours. Conclusions: Self-efficacy appears to be an important psychosocial construct that may directly or indirectly affect health behaviour to control diseases. Self-efficacy may also function as a link between effective health promotion and educational interventions and health behaviour change in disease control.


Introduction
The main aim of health promotion and educational interventions is to increase knowledge, attitudes and change behaviours to control or prevent diseases.According to health promotion interventions, knowledge and positive attitudes usually increase (Brega et al., 2013;Ezedinachi et al., 2002;Pai, Hong, & Hsu, 2006).However, sustained behaviour change is harder to achieve (Schwarzer, 2008).Knowledge alone may not always be a good predictor of improved health behaviour.The relationship between knowledge, attitude and behaviour is complex and difficult to establish (Chaffee & Roser, 1986).In order to understand this complex relationship, there are multiple theoretical perspectives that have been used to develop effective health promotion and educational interventions.Self-efficacy is seen as one such relevant psychological variable that connects the impact of health promotion and educational intervention on knowledge and attitude to health behaviour enactment (Albert Bandura & Wood, 1989;Bouffard-Bouchard, 1990;Strecher, 1983).

Self-Efficacy
Self-efficacy is defined as people's belief in their capabilities to achieve different levels of performance attainment (Bouffard-Bouchard, 1990).It is a key construct within several theories that are popular within health psychology such as Social Cognitive Theory (Albert Bandura, 1997), Protection Motivation Theory (Rogers, 1975), Transtheoretical Model (Prochaska & DiClemente, 1982), Health Action Process Approach (Schwarzer, 2014) and Theory of Planned Behaviour (A Bandura, 1977;Albert Bandura, 1978).According to Bandura in his Social Cognitive Theory, self-efficacy has an influence on preparing for action because self-efficacy levels can enhance or impede motivation to behaviour change, for instance 'I intend to reduce my smoking habit' or 'I intend eat fruit and vegetables regularly' (Maibach, Flora, & Nass, 1991).Apart from involving motivation, self-efficacy is also directly related to behaviour enactment, referring to the confidence that one can employ the skills necessary to resist temptation, cope with stress and mobilize one's resources required to meet situational demands (Schwarzer & Chung, 1996).

Mediator of Health Determinants
A large body of related research supports the predictive value that self-efficacy has on health behaviour change.Self-efficacy either has significant direct effect on behaviours (Baldwin et al., 2006;Hendriksen, Pettifor, Lee, Coates, & Rees, 2007;Oei & Burrow, 2000;Scholz, Sniehotta, & Schwarzer, 2005;Strauss, Rodzilsky, Burack, & Colin, 2001) or it becomes a mediator between other psychological variables to health behaviours (Darker, French, Eves, & Sniehotta, 2010;Motl et al., 2002).Although self-efficacy is an important determinant of a number of health behaviours, it is unclear of how to change this psychological construct as to give impact on behaviours as a result of health promotion and educational intervention (Ashford, Edmunds, & French, 2010).Currently, most reviews that were conducted usually involved interventions without health promotion methods and activities on effectiveness of self-efficacy as a mediator to physical activity or substance abuse and HIV prevention (Baranowski, Anderson, & Carmack, 1998;Hyde, Hankins, Deale, & Marteau, 2008;Lewis, Marcus, Pate, & Dunn, 2002;Marcus, King, Clark, Pinto, & Bock, 1996;Mize, Robinson, Bockting, & Scheltema, 2002;Sherwood & Jeffery, 2000).Hence, this review was carried out to identify interventions that applied health promotion and education activities as a tool to enhance self-efficacy for behavioural change in control and prevention of any related diseases.

Search Strategy
A comprehensive search strategy was developed to search all published articles that can be considered eligible for inclusion in this review.The search strategy was adapted from PubMed for use in the electronic databases.The literature search identified potential studies in all languages.Non-English language papers were translated and assessed for potential inclusion in the review as necessary.The following electronic databases were searched to identify potential studies: Full details of the search strings are given in Table 1.No restrictions were made regarding publication year.Two authors independently screened titles and abstract for inclusion in the review.Any disagreement was resolved by discussion between all authors.The screening process yielded 52 potentially relevant articles.Full text articles were retrieved and independently assessed by two authors.All disagreements were referred to third author., with suffix.mp (mp=ti, ab, sh, hw, tn, ot, dm, mf, dv, kw, ps, rs, nm, ui) Self-efficacy terms: Self-Efficacy OR Efficacy OR Self-Esteem OR Esteem OR Self-

Study Selection and Eligibility Criteria
A title and abstract review of these studies were conducted based on inclusion and exclusion criteria developed by the authors.

Inclusion Criteria
Participants for the included studies were not limited to specific characteristic such age or educational status.Only studies that used health promotion and education as intervention to prevent any diseases and measure the association between self-efficacy and behaviour change with quantitative outcomes were included in this review.

Exclusion Criteria
All studies based on laboratory or semi-field settings and qualitative analysis were excluded.Studies that focused on general health were also excluded in this review.

Validity Assessment
The risk of bias for included studies were assessed using 4 domains namely selection of participants, measurement of self-efficacy, analysis method used in assessing relationship between self-efficacy and health behaviour and finally attrition bias particularly loss to follow-up, missing data, or incomplete reporting (Table 2).Questionnaires that was translated from English to Spanish, scales of strongly agree to strongly disagree did not have exact equivalents in Spanish; so the validity of those data may have been compromised.

No adjustment
Relied on self-report of measurement of exercise behaviour.
In both structural model, relatively weak direct influences on the exercise behaviours, with path coefficients ranging from -0.04 to 0.23.Sharma SV., et. al., 2010 USA Not mentioned.Study of secondary data analysis.

Path analysis using LISREL
No adjustment.

Relied on secondary data
Cross-sectional study, causality cannot be determined.
Ethic homogeneity of the sample; limit generalizability.
No covariates were not adjusted for in the analysis Chariyeva Z., et. al, 2011 USA Of 183 participants, 85% were retained at 4-month follow-up, 78% at 8 month and 66% at the 12 month follow-up.

Multivariate analysis, mediation analysis
Age, educational attainment, having a main partner, sexual identity, date of HIV diagnosis, having an undetectable HIV viral load, clinic site, counsellor, proportion of counselling sessions done by telephone, history of previous enrolment in a study providing MI counselling, binge drinking in the last 3 months and cocaine/crack use in the last 3 months.
Self-reported data rather than observational.
Absence of randomization of counselling time; limit assessment on causal relationship between amount of motivational interviewing received and sexual behaviour.
On-going health promotion campaign during survey; influence participants to practice safer sex.
The study site was not randomly selected; threats to external validity and limit generalizability.

Data extraction
Relevant papers were retrieved and automatically entered into reference software (Endnote version 17).Data were extracted using a standardized data extraction form developed by the authors.Two review authors extracted study characteristics from included studies which include information on study design, total duration of study, study location, study setting and study period.Information on participants were also collected regarding total number of participants, mean age or age range and gender.For interventions, all information on description of intervention, duration and content of intervention were collected.All out comes were described as effectiveness of self-efficacy and its mediating effect on health behaviour.

Results
The search identified 314 citations of which 13 papers were included in the analysis.Figure 1 shows the flow diagram detailing the search process and inclusion of studies in this review.Due to significant heterogeneity of study designs, settings, participants, diseases, type of health promotion and education methods or activities and outcome measures, the key study outcomes were synthesized narratively according to health behaviour to control communicable and non-communicable public health diseases.Details of each study and overall validity across the entire set of included studies are broadly summarized in Tables 2 and 3.

Characteristics of Included Studies
Out of 13 papers included in this review, 5 papers are based in the United States of America, 2 in the Europe, 2 in South East Asia, 2 in Australia and the remainder 2 papers are from Africa and South America.Sample size ranged from 65 to 1113 and participants' mean age ranged from 9.37 to 62.5 years (Table 3).A one unit increase in self-efficacy was associated with a 52 and 39% decreased in unprotected sexual intercourse activity.
Self-efficacy was a mediator between number of counselling sessions and dose of unprotected sexual intercourse.
Apart from variation in the type of disease, settings, participants and educational component, the included studies also presented different types of health behaviours to change, which were promoted during the intervention.Most of studies were experimental in nature (9 papers) and 3 studies were observational with one study being primarily descriptive.The exact mix of interventions varied between studies.Four different approaches were used in the study designs: 1 study used an historical control period, measuring outcomes in the same setting at baseline and at a later time point, 2 studies included a control arm with no additional treatment as well as an intervention arm, 6 studies included a control arm exposed to some health promotion activity, along with an intervention arm and 4 studies are categorized as one-off survey following health promotion and educational intervention.All studies used questionnaires (4 administrated and 9 self-administrated) and other behaviour-related tool to measure the outcome.To measure self-efficacy, all studies used 5-points Likert Scale format questionnaires which were either pre validated or adapted from previous studies.However, there are a variety of tools used to measure health related behaviour; larval survey form, diet recall form or food record, physician prescription of lifestyle change, Block 2005 Food Frequency, pedometer, children physical activity questionnaires (CPAQ), calcium, osteoporosis and physical activity questionnaires (COPA), physical activity recall (3DPAR) and computer-assisted self-interview (ACASI).Most of these health behaviour tools of measurement were in self-reporting format.

Study Validity and Limitations
The potential limitations and factors that may influence validity of each individual study are described in Table 2.Among the issues affecting the overall dataset are the studies that relied on self-reported, and therefore prone to reporting bias, measures on eating and physical activity behavior (Chariyeva et al., 2013;Critchley, Hardie, & Moore, 2012;Luszczynska, 2004;Piaseu, Schepp, & Belza, 2002;Strathdee et al., 2009).Selective sampling in recruitment to a particular group of participants (Ievers-Landis et al., 2003;Piaseu et al., 2002;Sharma et al., 2010) or being geographically limited thus reducing generalizability of the study.
Attrition bias resulted from participants being lost to follow-up or the ability to complete the measurement of self-efficacy and/or health behaviour change.The response rate varied from 61% to 97%, but there are 2 studies that did not clearly highlight the issue (Sharma et al., 2010;Therawiwat, Fungladda, Kaewkungwal, Imamee, & Steckler, 2005).Multivariate analysis was the main analysis method performed by most studies.Four studies did not have any statistical adjustment because either they compared the intervention and control group using t-test (L.E. Burke, Dunbar-Jacob, Orchard, & Sereika, 2005) or they analysed the data for mediating and moderating effects or path analysis (Fuchs, Seelig, Göhner, Burton, & Brown, 2012;Luszczynska, 2004;Sharma et al., 2010).Demographics like age, ethnic, sex and education level are among the most factors included for statistical adjustment.
Another important issue here is selective outcome reporting.Not all the results may have been made available in the published version even though the studies measured many outcomes in wide range.Many studies were reported in different ways as categorical or continuous outcome which made it incomparable to assess the findings.

Self-Efficacy and Communicable Disease Control
One study investigated the effect of self-efficacy on dengue survey practices following community-based health promotion interventions in Thailand (Therawiwat et al., 2005).The programme was associated with a reduction in entomological indices of mosquito breeding and also with increased knowledge, perception and self-efficacy.However, in a multiple predictor model self-efficacy did not predict larval survey practices (p>0.05).
Three studies evaluated the impact of self-efficacy on behaviours related to control or prevention of HIV/AIDS transmission, two cross-sectional studies (Berendes & Rimal, 2011;Strathdee et al., 2009) and one two-armed randomized controlled trial study (Chariyeva et al., 2013).The first study evaluated self-efficacy as a facilitator of uptake of HIV testing behaviour following a communication-based behaviour change programme in Malawi (Berendes & Rimal, 2011).In both adjusted and unadjusted analysis, self-efficacy which was measured based on the individual's efficacy to remain abstinent (if not in a relationship), to remain faithful to one's partner, and to discuss and insist on use of condom was significantly associated with the previous testing behaviour or if a person had not yet been tested, with the desire to be tested.Self-efficacy was significantly associated with the willingness to get HIV tested (p<0.001).The second is study determined whether self-efficacy and injecting drug user status predicted condom use behaviour among Mexican female sex workers (Strathdee et al., 2009).A greater increase in self-efficacy was associated with greater condom use (p<0.05)regardless of injection drug use status.The third study assessed the mediation effect of self-efficacy and motivation on safer sex play practices among 490 People Living with HIV/AIDS (PLWHA) (Chariyeva et al., 2013).Self-efficacy was negatively associated with risky sexual practices.A one unit increased in self-efficacy was associated with 39% decreased in unprotected sexual intercourse activity.Moreover, self-efficacy was a mediator between number of counseling sessions and degree of unprotected sexual intercourse.
In a study of tuberculosis, self-efficacy was associated with medication-taking behaviour to control transmission (Morisky et al., 2001).This randomized study showed that self-efficacy strongly correlated with the completion of care (p<0.001).Peer counseling had a significant impact on self-efficacy for medication taking and on medication-taking-behaviour.

Self-Efficacy and Non-Communicable Disease Control
Seven studies examined the relationship between self-efficacy and behaviour related to healthy eating and/or exercise or physical activities to non-communicable control diseases; three studies were on osteoporosis prevention (Ievers-Landis et al., 2003;Piaseu et al., 2002;Sharma et al., 2010) and the rest are on coronary heart disease (L.E. Burke et al., 2005) hypertension, type-2 diabetes (Critchley et al., 2012) 3 and chronic orthopaedic conditions like arthritis (Fuchs et al., 2012).There is also one study related to breast self-examination behavior (Luszczynska, 2004).
The first study assessed self-efficacy using a osteoporosis self-efficacy scale (OSES) to measure confidence in women aged 21 years or younger to perform osteoporosis risk reduction behaviours focusing on exercise and dietary calcium intake (Schwarzer, Luszczynska, Ziegelmann, Scholz, & Lippke, 2008).After intervention, self-efficacy was found to directly predict calcium intake behaviour (p<0.05) and exercise (p<0.01).However, calcium intake was indirectly predicted by knowledge through health belief attitudes and self-efficacy.Path coefficient and R 2 statistics of causal analysis showed that self-efficacy was a mediator between attitudes and calcium intake behaviour (p<0.05), which accounted for 30% of variance as well as between knowledge and exercise (p<0.001), which accounted for 45% of the variance.
The second study on osteoporosis was a cross-sectional study among middle school girls to identify pathways by which psychosocial factors influence calcium intake and bone quality (Sharma et al., 2010).Confirmatory factor analysis and path analysis showed that self-efficacy did not predict calcium intake (p>0.05).However, self-efficacy had a significant indirect effect in calcium intake behaviour with mediation of outcome expectations (p<0.05).This indirect influence could be because participants with a higher confidence toward choosing calcium-containing foods could have higher outcome expectations, which could lead to higher calcium intake practices.
The third study regarding prevention of osteoporosis was carried out among 354 girls of a scout troop ages between 8-11 years old (Ievers-Landis et al., 2003).This study tried to predict the influence of self-efficacy and exercise self-efficacy based on Bandura's social cognitive theory.Even though self-efficacy significantly predicted dietary intake of calcium in presence of family support (p<0.001), it was not a significant predictor of frequency of exercise activity (p>0.05).However, a structural equation model found that, self-efficacy for calcium intake functioned as a partial mediator in the relationship between family support for exercise and calcium intake behaviour.
A randomized controlled trial investigated self-efficacy using cholesterol lowering diet self-efficacy scale (CLDSES) and adherence to a cholesterol lowering diet behavior (L.E. Burke et al., 2005).Participants were patients with hypercholesterolemia selected from three academic medical centre practice and research sites.Participants were given low cholesterol diet and/or pharmacological therapy for at least 2 weeks.Adherence behaviour was examined by comparing actual intake of total and saturated fat and cholesterol to prescribed amounts.Result showed that baseline self-efficacy significantly associated with concurrent adherence scores within the intervention group for total fat (p<0.05),saturated fat (p<0.01) and cholesterol (p<0.05).However, the control group only intake of cholesterol diet significantly associated with baseline self-efficacy (p<0.05).After intervention, the only adherence scores that were inversely related to self-efficacy were saturated fat for the intervention group, and saturated fat and cholesterol for the usual care group.The study assumed that self-efficacy was not a mediator of improving adherence behaviour to a low cholesterol diet in controlling coronary heart disease.
Critchley et al. conducted a randomized trial examining the psychological process of lifestyle change among 307 pre-diabetes adults following a six-session group-based intervention to promote healthier living (Critchley et al., 2012).They reported that, despite the fact that participation in the six small group's psychosocial-educational sessions was significantly associated with improvement in self-efficacy, it did not become a mediator between this participation and change in healthy diet and physical activity behaviour.Result also showed that knowledge about diabetes (p<0.05)significantly contributed to the effect of program participation on increased physical activity.However, none of those psychosocial variables significantly mediated change on eating behaviour.
A quasi-experiment longitudinal study on the cognitive mediation effects of a group-based intervention on exercise behaviour in 220 adults with chronic orthopedic conditions concluded that, self-efficacy did not directly predict exercise behavior (Fuchs et al., 2012).However, self-efficacy mediated the impact of intention strength and self-concordance.
Luszcynska et al. reported a cluster randomised trial for promotion of breast self-examination to reduce mortality from breast cancer amongst 417 female students in Poland (Luszczynska, 2004).This study concluded that maintenance self-efficacy predicted change in breast self-examination performance (p<0.05) as well as intention and planning in the intervention group.Pre-action self-efficacy however, only significant in predicting intention to do breast self-examination in both intervention and control groups.
Although the studies reviewed generally conceptualized self-efficacy as perceived capabilities of performing specific health behaviours, operational measures of self-efficacy varied considerably.A two-step approach was used to measure self-efficacy by determining a person belief on behaviour modification could be accomplished and to rate the strength of the belief (Albert Bandura, 1997).In many of the studies, this method was incorporated into a unified set of measures that included both strength and level of self-efficacy.In most studies, participants were asked how confident they would feel in performing targeted health behaviour in different situations where the ability to perform the behaviour might vary accordingly.
In studies where other psychosocial constructs like family support, attitudes, intention strength and self-concordance, as well as outcome expectations were examined, self-efficacy was found to be a distinct and powerful predictor for health behavior (Fuchs et al., 2012;Ievers-Landis et al., 2003;Piaseu et al., 2002;Sharma et al., 2010).Other findings suggest that it interact with or influence self-efficacy to change or maintain targeted health behaviour to control diseases.Self-efficacy may depend on perceived influence in controlling outcomes.This supports Bandura's Social Learning Theory that psychosocial arousal influences self-efficacy and self-efficacy reflect a person's perceived abilities rather than one's true abilities that often influence behaviour change and maintenance (A Bandura, 1977).
Perceptions of susceptibility to a particular illness and of benefits of taking recommended preventive actions could both be considered expectations of outcome.Outcome expectations are a major component of value-expectancy models of health behaviour such as the Health Belief Model (Becker, 1974;Janz & Becker, 1984).These expectations may play the largest role in influencing initial motivation and decision to change a health practice (Sharma et al., 2010).Health practices which are not particularly difficult to modify but whose perceive consequences are uncertain may largely depend on outcome expectations like compliance to treatment in tuberculosis (Morisky et al., 2001).Studies have found that outcome expectations, such as perceived susceptibility to tuberculosis-related illness or perceived benefits of taking the medication, are associated with compliance to tuberculosis treatment (Morisky et al., 2001).
In situation where belief and trust play a major role among sexual worker, self-efficacy is inadequate to mediate the safe health practice in using condoms to prevent sexually transmitted diseases (Chariyeva et al., 2013;Strathdee et al., 2009).Finally, both outcome and self-efficacy may be required to explain behaviour change process when there is a need for a change in practices of uncertain consequences.This might be applied during the modification of certain dietary habits, such as reducing the intake of saturated fat to reduce risk of hypertensive complications like stroke (V.Burke et al., 2007) or reducing foods that high in cholesterol to reduce risk of coronary heart disease (L.E. Burke et al., 2005).
The findings of this review support the perspective of Bandura's Social Learning Theory regarding the concept that self-efficacy relates to beliefs about abilities to change specific behaviours in particular situations (Schunk & Carbonari, 1984).Self-efficacy does not refer to a personality characteristic or a global trait that operates independently of contextual factors (A Bandura, 1977;Albert Bandura, 1978).The individual's self-efficacy will vary greatly depending on the particular task and context.Therefore, it is inappropriate to characterize an individual as having 'high' or 'low' self-efficacy without reference to the specific health behaviour and circumstance with which self-efficacy judgment is associated, i.e. control or prevent any particular diseases.The finding of this review also supports the previous studies on the predictive value of self-efficacy has on health behaviour change either direct effects or as an effective mediator between other psychological variables with health behaviours (Baldwin et al., 2006;Darker et al., 2010;Hendriksen et al., 2007;Motl et al., 2002;Oei & Burrow, 2000;Scholz et al., 2005;Strauss et al., 2001).

Conclusions
In conclusion, the majority of the studies identified in this review found a significant association between self-efficacy in both short and long term success of health promotion interventions.The investigators had undertaken the appropriate analyses to show the effect of self-efficacy on psychological factors and behavioural change.This review supports the hypothesis that the concept of self-efficacy is a good psychosocial construct that can translate knowledge and information from health promotion and educational intervention into behavioural action.Improving self-efficacy must be one of main agenda in successful behavioural-impact health promotion and educational interventions to control and prevent diseases.Future health promotion and educational intervention to control or prevent diseases must be designed to influence related health behaviour by giving more attention on enhancing self-efficacy implicitly.

Ethics Approval and Consent to Participate
Since only official public data were used in this study, neither institutional board approval nor informed consent was required.

Figure 1 .
Figure 1.Search process and study selection

Table 1 .
Search String Search Strategy for Ovid SP MEDLINE and EMBASE Efficacy of Social Cognitive Theory OR Self-Efficacy of Health Belief Model OR Self-efficacy of Health Psychology

Table 2 .
Validity Assessment and Limitations

Table 3 .
Study Characteristics and Outcomes The topics covered by the studies included dengue survey practices (one study), medication taking behaviour (one study), breast self-examination behavior (one study), safe sex and HIV test behaviour (one study), health eating with physical activity behavior (one study), eating behaviour only (one study) and physical activity(one study).Most papers (8 studies) are mainly on non-communicable diseases (breast cancer, coronary heart disease, osteoporosis, hypertension, and type-2 diabetes.The remaining 3 papers have a range of communicable diseases such as dengue fever, tuberculosis and HIV/AIDS.Studies varied with regard to types of health promotion and education intervention.Most of studies applied a group of educational activities including use of print or electronic media, group discussions or training sessions, counseling or meeting sessions, demonstrations or talks, individual advice and behavioural contract.