At its core, education and training initiatives provide continued education for nutrition professionals to alter behavioural intention^{(Reference Brown1)}. Behavioural intention describes the beliefs, assumptions and personal factors that influence a given behaviour^{(Reference Ajzen, Kuhl and Beckmann2)}. Certain theories postulate that factors such as attitudes, perceived behavioural control, and normative beliefs influence the likelihood of behavioral intention change^{(Reference Ajzen, Kuhl and Beckmann2)}. For example, the three theoretical constructs identified previously are the motivational factors from the Theory of Planned Behaviour (TPB) that determined the likelihood of specific behaviour change^{(Reference Ajzen, Kuhl and Beckmann2,Reference Montaño and Kasprzyk3)} . TPB assumes a causal linkage between normative and behavioural beliefs to intentions directed by attitudes, perceived behavioural control and subjective norms and has framed numerous health behaviour change studies and educational messaging^{(Reference Ajzen, Kuhl and Beckmann2,Reference Montaño and Kasprzyk3)} . TPB is applied to many and varying health behaviours studies that are previously reviewed and resulted positively for rigour and effectiveness^{(Reference McEachan, Conner and Taylor4,Reference Hagger, Cheung and Ajzen5)} . The prevalence of TPB usage in public health and its applicability to nutrition professional education is in part due to its ease of operationalisation in developing, analysing and measuring behaviours and interventions^{(Reference McEachan, Conner and Taylor4,Reference Hagger, Cheung and Ajzen5)} .

The capability to effectively predict and describe factors that influence behavioural intention is important to research in health including nutrition, physical activity and sex education and assists to develop interventions to change specific behaviours^{(Reference McEachan, Conner and Taylor4)}. For example, Asare studied condom use among college students using the TPB and reported behavioural intention significantly predicted condom use^{(Reference Asare6)}. Additionally, other interventions including communicating health messages to clients^{(Reference Lee, Brennan and Gibson7,Reference Fishbein and Cappella8)} , and health behaviour education interventions^{(Reference Asare6,Reference Steinmetz, Knappstein and Ajzen9,Reference Casper10)} , that utilised the TPB see changes in attitudes, perceived behavioural control, normative beliefs and intention among participants. Hence, it is necessary to understand the current knowledge, self-efficacy and attitudes of a study population such as nutrition professionals when designing an educational training to ensure that messaging is directing behavioural intention.

Unfortunately, health education and interventions today are not always theory-based, which can lead to misconceptions and limited dissemination of evidence-based information^{(Reference Brown1)}. Misconceptions, about behaviours due to inconsistent or lack of information, can lead to an absence of individual intention to participate in activities^{(Reference Lee, Brennan and Gibson7)}. Frequently, individuals do not have a clearly defined behavioural response to a particular information^{(Reference Miller and Oaks11)}. When a population has limited experiences or education with the intended behavioural change, then their attitudes, beliefs and knowledge must be shaped into a new behaviour^{(Reference Miller and Oaks11)}. This response-shaping process is important when new information is constantly evolving or there is a limited understanding or experience with the information within the message^{(Reference Miller and Oaks11)}, which is especially relevant for emerging science and updated professional practice. A potential response-shaping process is targeted, theory-based trainings for populations with low self-efficacy, attitudes and normative beliefs. For example, nutrition practitioners have a limited understanding and knowledge of Dissemination and Implementation (D&I science), which demands education and messaging to be formulated in a way that will change behaviour^{(Reference Koorts, Naylor and Laws12)}.

D&I science is the study of the integration and translation of research findings into health practices or interventions^{(Reference Brownson, Colditz, Proctor, Brownson, Graham and Proctor13,Reference Colditz, Emmons, Brownson, Graham and Proctor14)} . D&I science describes two components: dissemination, which is defined as the active approach of spreading evidence-based interventions to the target audience, and implementation, which is defined as the process of putting to use or integrating evidence-based interventions within a setting^{(Reference Rabin, Brownson and Haire-Joshu15)}. The historical roots of D&I science stem from the diffusion of innovations theory^{(Reference Rogers16)} and the agricultural extension model^{(Reference Dearing, Kee, Brownson, Graham and Proctor17)}, which provided the building blocks for evidence-based public health research that led to D&I science foci in cancer, mental health and substance use^{(Reference Brownson, Colditz, Proctor, Brownson, Graham and Proctor13,Reference Dearing, Kee, Brownson, Graham and Proctor17)} . In recent years, specific fields in public health have increasingly recognised the need to use D&I science to enhance rigour of interventions (policies, programmes, education, information and knowledge translation). Furthermore, evaluations of current D&I trainings to increase capacity and usage of D&I science in health highlight improvements in knowledge, confidence and changes to research practice (incorporation of D&I science into research)^{(Reference Goodenough, Fleming and Young18–Reference Ramaswamy, Mosnier and Reed26)}. A recent review describes how more involvement in D&I-focused research has led to the evaluation of adaptations to programme implementation, which has increased the systematic documentation and reporting to improve replication and patient-centred outcomes^{(Reference Glasgow, Harden and Gaglio27)}. Yet, a recent systematic review highlighted the current contexts of D&I science trainings, in which zero were explicitly focused on nutrition^{(Reference Davis and D’Lima20)}. Without proper training, nutrition practitioners and researchers have an incomplete understanding about what D&I science is, which hinders their usage and involvement and impacts the translation of health information and interventions to populations^{(Reference Koorts, Naylor and Laws12,Reference Warren, Frongillo and Rawat28–Reference Tumilowicz, Ruel and Pelto31)} . Therefore, reviews suggest that nutrition-specific trainings are needed to increase the use of D&I science among nutrition practitioners and researchers to improve nutrition interventions, dietary patterns and nutrition-related outcomes^{(Reference Walker, Wattick and Olfert30)}. Without such training, nutrition interventions will continue to see implementation challenges such as the absence of sustained, effective and appropriately adapted nutrition educations, programmes or policies^{(Reference Koorts, Naylor and Laws12,Reference Walker, Wattick and Olfert30)} . Thus, the purpose of this research study was to determine how nutrition practitioners’ attitudes, behavioural control and normative beliefs influenced their intention to use D&I science in their professional practice to best develop and formulate a TPB-based nutrition-specific D&I training.

Materials and methods

IRB approval

This study was approved by BLINDED FOR REVIEW Institutional Review Board (protocol # BLINDED FOR REVIEW).

Theoretical framework

TPB (Fig. 1) was developed in the 1980s by Icek Ajzen using particular constructs to predict human behaviour^{(Reference Ajzen, Kuhl and Beckmann2)}. This theory postulates that attitude towards a given behaviour, normative beliefs about said behaviour and perceived control over that behaviour are all predictors of behavioural intention^{(Reference Ajzen, Kuhl and Beckmann2)}.

Fig. 1 Theory of Planned Behaviour. ^aShaded areas represent the theory of reasoned action. ^bFigure from: Montaño & Kasprzyk (2014)^{(Reference Montaño and Kasprzyk3)}

Behavioural intention, the first construct of the TPB, is defined as the motivational factor to achieve a specific behaviour^{(Reference Ajzen, Kuhl and Beckmann2)}. The stronger the intention, the more likely it is that an individual will complete the intended behaviour^{(Reference Asare6)}. The attitude towards the behaviour, the second construct of the TPB, describes someone’s positive or negative thoughts or reactions towards the given behaviour^{(Reference Ajzen, Kuhl and Beckmann2)}. The third construct of the TPB, normative beliefs, defines the social pressures or influences someone experiences about the given behaviour^{(Reference Ajzen, Kuhl and Beckmann2)}. The final construct, perceived behaviour control (i.e. knowledge, learning autonomy) refers to someone’s ability to and knowledge of a given behaviour and how that influences their capacity to perform^{(Reference Ajzen, Kuhl and Beckmann2)}.

Participants and recruitment

This study was a cross-sectional study to identify current attitudes, perceived behavioural control and normative beliefs among nutrition practitioners (including clinical and private practice dietitians, researchers, outpatient nutrition educators, nutrition and dietetic students, academics (professors, researchers)) to understand what factors would most likely assist in increasing intention to use D&I science. To recruit participants, a weekly email was sent to the Society for Nutrition Education and Behaviour membership listserv (referred to as SNEEZE), which includes roughly 1000 members from a variety of different occupancies such as clinical dietetics, higher education, private sector and nutrition educators at a variety of locations such as State Cooperative Extension, public health department, private practice clinics and outpatient settings. An initial email was sent to the SNEEZE listserv in early September 2020 containing a Qualtrics survey link (Qualtrics, Provo, UT, 2021). A reminder email was sent every 2 weeks until the beginning of November 2020. The survey remained active from September 2020 to November 2020.

Hypothesis testing

For non-parametric testing, four hypotheses were tested and listed below.

1. 1. Attitudes toward behaviour (use of D&I science), subjective norms, descriptive norms, injunctive norms (described as normative beliefs) and perceived behavioural control will be positively correlated with intention.

2. 2. Higher expert score (proficient and expert level) will be positively correlated with attitudes, normative beliefs and perceived behavioural control.

3. 3. Training recipients (having received training) will be positively correlated with higher perceived behavioural control.

4. 4. Perceived behavioural control will be positively correlated with attitudes and normative beliefs.

Evaluation survey

The questionnaire was constructed based on an existing TPB survey development manual^{(Reference Francis, Eccles and Johnston32)} and validated measurements^{(Reference Fishbein and Ajzen33)}. The survey development manual was designed to assist researchers in health services to produce consistent and effective questionnaires to measure TPB constructs^{(Reference Francis, Eccles and Johnston32)}. These TPB constructs have been validated and used in a variety of different settings and populations to predict behavioural intention and ultimately create interventions to change behaviours^{(Reference McEachan, Conner and Taylor4)}.

Before accessing the online survey, participants were required to read and accept informed consent via two questions. Then, participants would complete twenty-four questions and one question to gather contact information for gift card recipients. One survey question assessed for previous training (‘have you had any prior training in Dissemination and Implementation science?’) and another survey item measured self-ranked D&I expertise (categories described below). The remaining twenty-two questions assessed the constructs of the TPB through a 7-point Likert scale (1 – strongly disagree and 7 – strongly agree) and open-ended questions. The behaviour in question was the use of D&I science. The survey was reviewed and approved by all authors prior to survey dissemination. Evaluation survey is provided as online supplementary material.

Measurement scoring

Expert ranking

Participants categorised their perception of D&I experience. Categories and definitions included inexperienced (never heard of D&I science before); novice (a person new to or inexperienced in the D&I field, i.e. limited understanding of D&I science); beginner (a person just starting to learn D&I science, i.e. have some knowledge base, e.g., could vaguely define D&I science); proficient (competent or skilled in doing or using something, i.e. have conducted D&I research and/or attended webinars/trainings/conference sessions and could explain to others the many attributes of D&I science) or expert (displaying special skill or knowledge derived from training or experience (i.e. have numerous experiences conducting D&I research and/or attended many webinars/trainings/conference sessions and could confidently explain/teach others the many attributes of D&I science). Responses were coded as ordinal variables (with five distinct levels) for quantitative analysis.

Participant training

For training scores, participants were asked to describe if they had ever received D&I training in the past. Qualitative responses were manipulated to binary categorical variables (yes or no) for analysis. Additionally, descriptions of received D&I trainings were grouped to provide further information.

Theory of Planned Behaviour constructs score

Each construct (i.e. attitudes, normative beliefs (subjective, descriptive, injunctive), perceived behavioural control and intention to use) was developed and measured based on previous validated TPB survey tools and measurements^{(Reference Francis, Eccles and Johnston32,Reference Fishbein and Ajzen33)} . Each survey question directly measured each construct. Survey responses to each construct were averaged for an overall mean score and were used as continuous variables for quantitative, non-parametric analysis.

Analysis

Quantitative data were analysed in JMP software (version pro 14, SAS Institute Inc.) and included descriptive statistics, frequency analysis and non-parametric measures. Non-parametric measures are statistical methods in which the data are not assumed to come from prescribed models that are determined by a small number of parameters or do not meet all statistical assumptions needed for parametric testing. This data did not meet all statistical assumptions for parametric testing (e.g. multivariable regression). Therefore, Spearman’s ρ tests were used to determine the strength and direction of correlations between TPB measures to understand relationships between TPB constructs. Kruskal–Wallis tests can determine how categorical independent variables relate to a continuous variable; however, this test cannot tell you which specific groups are statistically significant (an omnibus test statistic). For this study, Kruskal–Wallis tests were used to understand whether TPB measures (continuous variables) differed based on expert ranking and training. Since Kruskal–Wallis tests are an omnibus test statistic, Steel–Dwass tests were used to pairwise compare categories in expert rankings and participant training with TPB measures.

Results

Demographics

Table 1 describes descriptive statistics from quantitative analysis. The demographics of the participants (n 70) are similar to the clinical dietetics field^{(Reference Robinson34)}, yet they are not entirely representative or generalisable of all nutrition practitioners. Additionally, categorised open-ended questions on individual’s job or place of work resulted in most respondents worked in academics (n 31) as a professor, researcher or student. Many participants worked within outpatient nutrition education interventions (n 18), and other survey answers included clinical dietitians (n 9) (in-patient, outpatient and private practice) and administration (n 4). Furthermore, respondents could elect to describe the variety of D&I science trainings they received. In which, many (n 15) described receiving training during graduate school seminars, national conferences (n 7), grant writing workshops (n 12) and work experience (n 4) such as hospital administration or Cooperative Extension. Only two participants reported receiving explicit D&I science training including the Training Institute for D&I Research in Cancer or Training Institute for D&I Research in Health^{(Reference Meissner, Glasgow and Vinson35)}.

Table 1 Participant demographics

Non-parametric measures

Table 2 reports quantitative analysis from the Spearman’s ρ analysis to test observed correlations among TPB variables. Significant positive correlations were shown between several independent and dependent variables including perceived behavioural control and intention score (P = 0·0119) and attitude (P = 0·0074); perceived behavioural control and subjective normative belief (P = <0·0001), injunctive normative belief (P = 0·0017) and descriptive normative belief (P = 0·0006).

Table 2 Correlations between TPB measures among participants

TPB measure, Theory of Planned Behaviour measures; Prob > |ρ|, probability of obtaining a Spearman’s ρ value greater than the one shown.

* Signifies statistical significance.

Table 3 reports quantitative analysis from the Kruskal–Wallis test analysis to test observed correlations among variables. Significant positive correlations were shown between attitudes and training (P = 0·0022); perceived behavioural control and expert score (P = 0·0126) and attitudes and expert score (P = 0·0025).

Table 3 Correlations between TPB measures, participant training and expert ranking

TPB measure, Theory of Planned Behaviour measures; Prob > ChiSq, probability of obtaining a chi-square value greater than the one shown.

* Signifies statistical significance.

Table 4 reports the pairwise comparisons from the Steel Dwass. A significant positive correlation was identified between a higher training score (receiving a training) and a higher attitude score (P = 0·0023). Additionally, the pairwise comparison suggests higher levels of expertise correspond to attitudes and specifically in the proficient group when compared with the inexperienced group (P = 0·0016). Lastly, higher levels of expertise correspond to higher perceived behavioural control score (P = 0·0445), specifically compared between proficient and beginner groups.

Table 4 Steel–Dwass test comparison between expert ranking, TPB measures and previous participant training

CL, confidence level; IE, inexperienced; TPB, Theory of Planned Behaviour.

* Signifies significance.