CVD is the leading cause of non-communicable deaths worldwide( 1 ). The direct and indirect costs associated with CVD are high, with CVD-related health-care costs accounting for 12 % of the total Australian heath-care budget in 2008–09 ($AU 7605 million) and an estimated lost income of $AU 1·1 billion due to exit from the labour force (2009)( Reference Schofield, Shrestha and Percival 2 , 3 ). As the population in Australia ages, so too does the economic burden of chronic CVD conditions( Reference Ha, Hendrie and Moorin 4 ). Nutrition is recognised as an important contributor to the prevention of primary and secondary CVD events( Reference Anderson, Gregoire and Hegele 5 – Reference Board 8 ). Dietary intakes affect the biochemical pathways contributing to hypercholesterolaemia, hypertension, hyperglycaemia, insulin resistance and inflammation, which contribute to the development and progression of CVD( Reference Mathews, Liebenberg and Mathews 9 ). High-quality diets, such as those containing greater amounts and variety of fruits and vegetables and lower amounts of energy-dense, nutrient-poor foods, are associated with a lower risk of subsequent CVD related morbidity and mortality in those with pre-existing CVD risk factors( Reference Dehghan, Mente and Teo 10 ). However, the Prospective Urban Rural Epidemiology study found that of 7519 individuals from seventeen countries who had experienced a self-reported CVD event, only 39 % were considered to have a healthy diet at 4–5 years following the event( Reference Teo, Lear and Islam 11 ). This indicates that appropriate nutrition knowledge for the prevention and treatment of CVD failed to be incorporated into long-term behaviour change for these individuals. It is likely that this was due to a range of reasons.
Knowledge translation (KT) describes the process that encompasses the stages from development and synthesis of the evidence-based knowledge through to the translation of this knowledge by health-care providers to consumers into subsequent health behaviours, with the end goal of improved individual health. The Canadian Institute of Health Research defines this process as ‘a dynamic and iterative process that includes synthesis, dissemination, exchange and ethically-sound application of knowledge’( Reference Tetroe 12 ). It is also described as the knowledge-to-action cycle where a knowledge creation process precedes an action cycle in which the created knowledge is utilised by a range of decision makers and stakeholders, ranging from patients to health-care policy makers( Reference Graham, Logan and Harrison 13 , Reference Straus, Tetroe and Graham 14 ). It is the action cycle that is the focus of the present review, where end users implement and utilise the evidence-based knowledge( Reference Straus, Tetroe and Graham 15 ). As this is a behaviour change process, it is suggested that implementation of KT should be based on a theoretical framework( Reference Eccles, Grimshaw and Walker 16 , Reference Graham and Tetroe 17 ).
In a previous review (2012) of KT in the allied health fields of nutrition and dietetics, occupational therapy, pharmacy, physiotherapy and speech pathology, Scott et al.( Reference Scott, Albrecht and O’Leary 18 ) found that research publications reported mixed results from studies seeking to translate knowledge into practice. Studies of KT strategies were reported generally to be of poor methodological quality and no particular type of KT strategy was shown to be more effective than others. Education only as a KT strategy was commonly employed, with consistent non-significant results. A literature review of strategies used to achieve lifestyle changes following CVD events found that education was commonly used to support adherence to heart-healthy dietary recommendations, with staff highly trained to provide the intervention( Reference Cobb, Brown and Davis 19 ). However, that review did not focus on diet exclusively, nor how nutrition knowledge was translated. It is the need for this nutrition KT that was specifically highlighted in the European Guidelines on CVD prevention in clinical practice (2012): ‘The challenge for coming years is to translate nutritional guidelines into diets that are attractive to people and to find ways in which to make people change their (long-standing) dietary habits’( Reference Perk, De Backer and Gohlke 7 ).
The objective of the present systematic review was to identify how the best available current evidence on diet for the prevention and treatment of CVD is translated into practice in those with or at high risk of CVD. The primary aim was to identify aspects of successful health-service nutrition translation studies in CVD in terms of the methodology, including theoretical framework, implementation strategies, programme design, resources, use of technology and message transmission channels. A secondary aim was to evaluate the methodological quality of these translation studies and the effectiveness of nutrition evidence translation on diet-related CVD risk factors.
The conduct and reporting of the present systematic review adhered to the guidelines stated in the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) Statement( Reference Moher, Liberati and Tetzlaff 20 ). The systematic review protocol was registered with PROSPERO (http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42014007404) as CRD 42014007404.
Adults (classed as 18 years or older) with one or more CVD diagnoses were included. Relevant CVD diagnoses included angina (stable or unstable), coronary artery disease, CHD, myocardial infarction, acute myocardial infarction, or intervention such as coronary artery bypass graft or percutaneous transluminal coronary angioplasty. In cases where adults and minors were included together, results for adults had to be reported separately. Due to the nature of some CVD where living skills may be diminished, interventions targeting the carers of people with a CVD diagnosis were also included. As KT studies can also relate to service providers, health professionals directly treating those with a CVD diagnosis were included, as were interventions targeting heath-care systems or policies directly related to the treatment of patients with diagnosed CVD conditions. The review also intended to provide KT strategies for nutrition in the treatment and prevention of CVD; thus those defined as high risk, 16 % and above by the 5-year absolute CVD risk assessment( 21 ), were also included. If an intervention population was a combination of those at moderate and high risk of CVD, either a minimum of 50 % of the population had to be at high risk or the results had to be reported separately.
Study inclusion criteria
Studies were included with either experimental or quasi-experimental designs, with comparators, control groups, wait-list control groups or pre–post designs. Studies were limited to English language in the years from 1985 to 2013, as there is no evidence to show that a lack of studies in languages other than English will bias the results( Reference Morrison, Polisena and Husereau 22 ), and the time span reflects recent acknowledgement of the need for translation strategies (see online supplementary material, ‘Additional File 1: Search strategy and results for the systematic review’). To be relevant for inclusion, studies needed to assess and report diet or nutrition KT as a separate measureable outcome.
The methods of the published study interventions were required to clearly state the dietary or nutrition knowledge that was to be translated and the method by which this KT was to occur. The KT needed to be applicable to one or more of the following areas: evidence synthesis, dissemination, exchange, application or ethically sound application of knowledge. No set length of follow-up was determined as this would be dependent on the intervention delivered. As the focus of the review was on diet and nutrition, the KT had to relate to whole foods, not nutrient supplements only. For example, KT regarding fish intakes would be considered eligible, whereas a focus on n-3 fatty acid supplementation would not. Studies including supplementation within their intervention were not excluded if whole foods were reported separately.
The KT could be at the personal, community or health-care level. Both pre- and post-intervention outcome measures had to be reported in the results to evaluate effectiveness.
The primary outcomes were dependent upon the stage of the KT spectrum at which the intervention was determined to occur. Outcomes for KT at patient, caregiver, health professional, health system and public health levels were the following.
1. Patient level: cardiovascular risk markers, e.g. serum lipids, blood pressure, arterial stiffness, anthropometrics.
2. Caregiver level: cardiovascular risk markers, e.g. serum lipids, blood pressure, arterial stiffness, anthropometrics.
3. Health professional level: changes in practice.
4. Health system level: changes in cardiovascular prevention or treatment policies, guidelines, recommendations or best practice.
5. Public health or community level:
a. improvement in cardiovascular risk rates, as measured by biochemical risk markers;
b. hospitalisations, morbidity or mortality due to CVD; and
c. health expenditure per capita.
Secondary outcomes appropriate to the stage of KT were the following.
6. Patient level: knowledge or behaviour change related to dietary intake.
7. Caregiver level: knowledge or behaviour change related to dietary intake.
8. Health professional level: none.
9. Health system level: none.
10. Public health or community level: none.
Search strategy and selection of studies (information sources)
A search strategy was developed and implemented with the search conducted in the databases of CINAHL, Cochrane, EMBASE, MEDLINE, PsycINFO and Scopus. Key terms for KT were sourced from Armstrong et al.( Reference Armstrong, Waters and Dobbins 23 ) and Scott et al.( Reference Scott, Albrecht and O’Leary 18 , Reference Scott, Albrecht and O’Leary 24 ) and cardiovascular terms were identified from two Cochrane systematic reviews by Hooper et al.( Reference Hooper, Bartlett and Davey 25 ) and Hartley et al.( Reference Hartley, Flowers and Lee 26 ). Nutrition terms encompassed nutrients and nutrition, eating, foods, diets and terms specific to CVD such as dietary fats. Reference lists from included and related studies, relevant conference abstracts and theses were searched for additional citations. Protocol publications or references to gain further information on methods related to the included studies were sourced and included if relevant. Multiple publications from the same intervention were combined and all relevant outcomes reported.
Process of study selection
Two reviewers independently assessed records based on title and abstract for eligibility and full text retrieval. Full-text articles were assessed independently by two reviewers. Disagreements were discussed and until consensus was reached, with a third reviewer consulted when consensus was not reached.
Data were extracted using a spreadsheet, initially piloted for consistency and to ensure all required data were obtained. One reviewer extracted all of the data and a second reviewer checked the extracted data for accuracy and consistency. Disagreements were discussed until consensus was reached. Data items included for extraction related to details about the population, intervention, use of control groups and study outcomes related to nutrition or diet. Details regarding the KT strategy used were also extracted, such as the framework, theory or principle on which the translation strategy was based, the behaviours targeted and the change techniques used, as defined by Abraham and Michie( Reference Abraham and Michie 27 ).
Assessment of study quality
Risk of bias in individual studies was assessed by two reviewers independently using a tool adapted from the American Dietetic Association Evidence Analysis Manual quality criteria checklist( 28 ) and relevant items from the checklist of the Workgroup for Intervention Development and Evaluation Research (WIDER) recommendations( Reference Albrecht, Archibald and Arseneau 29 ). These items included a detailed description of the intervention, clarification of the assumed change process and access to intervention manuals( Reference Albrecht, Archibald and Arseneau 29 ). The combined tool (see online supplementary material, ‘Additional File 2: Quality checklist’) was required to assess both the quality of the nutrition study and also of the behaviour change intervention. Each item was coded as Yes (clearly indicated and present), No (missing or not appropriate), Unclear (if indicated, but insufficient information provided) or Not Applicable (N/A; see ‘Additional File 2: Quality checklist’ for questions to which N/A could be applied). Each item was considered individually. Relevance questions 1–4 (see ‘Additional File 2: Quality checklist’) and validity questions B, C, F and G were weighted for importance according to the American Dietetic Association Evidence Analysis Manual ( 28 ) and quality for the items from WIDER recommendations( Reference Albrecht, Archibald and Arseneau 29 ) was reported separately. Quality items as per the WIDER recommendations were given as questions K, L and M. Quality was designated as positive if both questions K and L were yes, neutral if either K or L was positive and M was positive, and negative for one or zero yes in total for all WIDER questions. K and L were designated as the more important translation items as clinicians would have access to their own resources as appropriate (question M).
Meta-analysis of data was not expected to be possible due to heterogeneity in both the stage of KT in which interventions could occur and the expected reporting of outcomes. Data were synthesised into the study characteristics, KT characteristics, intervention content and study quality by one reviewer, with data synthesis checked for accuracy and consistency by a second reviewer. Data were further stratified by target population: (i) those with CVD diagnosis/diagnoses; (ii) populations assessed as being at high risk of CVD; (iii) interventions targeting heart failure patients; and (iv) interventions targeting health professionals. Data were stratified using these methods because:
∙ those with a CVD diagnosis would be expected to have different perceptions of CVD risk or may be at a different stage of change compared with those assessed at high risk of CVD;
∙ those with heart failure would be expected to receive dietary advice primarily addressing sodium and fluid restrictions; and
∙ interventions targeting health professionals would be expected to use a substantially different theoretical framework from those perceiving risk to self.
The search strategy identified 4420 titles after duplicates were removed (see Fig. 1), with 354 full texts retrieved and screened for inclusion/exclusion, and forty-three texts included that described thirty-five separate studies( Reference Aish 30 – Reference Van Der Weijden, Grol and Schouten 72 ). Primary reasons for exclusion were populations not specifically related to CVD and outcomes not related to nutrition KT.
Risk of bias within studies
The quality of the studies was assessed using the composite tool described above and provided in the online supplementary material, ‘Additional File 2: Quality checklist’. Three major areas of quality were assessed as relevance, internal validity and strength of KT reporting. The results of the quality assessments are reported in ‘Additional File 3: Methodological quality scores and risk bias assessment in nutrition knowledge translation studies for cardiovascular disease’. No studies were excluded based on quality, as all studies were determined to have limitations in at least one major area related to study quality. Five studies were scored negatively for relevance, due to lack of feasibility of study replication with a limited budget, with the remaining twenty-nine studies assessed as positive in terms of relevance. Seven studies achieved a positive rating for internal validity, with twenty scored as neutral and eight scored as negative. There was limited reporting in terms of details related to the translation of nutrition and dietary knowledge as only three studies were scored positive, three scored as neutral and twenty-nine scored negative. No studies scored positive responses in all three areas, although four studies scored two positive and one neutral response( Reference Allen 32 , Reference Goodwin, Forman and Herbert 45 , Reference Luszczynska, Scholz and Sutton 49 , Reference Powell, Calvin and Richardson 66 ). Three scored neutral for validity( Reference Allen 32 , Reference Goodwin, Forman and Herbert 45 , Reference Luszczynska, Scholz and Sutton 49 ) and one for KT strategy( Reference Powell, Calvin and Richardson 66 ). Three studies scored negatively or neutral in all three areas( Reference Gleason, Bourdet and Koehn 44 , Reference Hofman-Bang, Lisspers and Nordlander 46 , Reference Sundin, Lisspers and Hofman-Bang 54 ). In particular, these three studies all scored negatively for relevance as they were judged as not being feasible for a clinician to replicate in the context of current cardiovascular clinics due to high resourcing costs, although they may have been feasible during the time at which the studies were performed. Excluding relevance, these studies were not significantly different from many of the other studies included in the review in either the way KT was reported or the level of detail provided to determine validity.
Disagreements between evaluators of manuscript quality were found in five of the thirty-five reviewed manuscripts, in individual questions contributing to the overall categories of validity and KT strategy. However, these did not affect the overall category grading given for these manuscripts.
Although all included studies were evaluated as translating nutrition-based knowledge as part of their intervention strategy, no studies specifically stated nutrition or dietary KT to be their primary or secondary aim. Therefore the internal validity of the KT strategy cannot be established as it cannot be proven that it was directly responsible for the primary and secondary outcomes. However, as KT was a fundamental part of the intervention strategy, the effectiveness of the translation has been inferred through the measures used.
In total, thirty-seven publications from thirty-one separate studies targeted patients( Reference Aish 30 – Reference Powell, Calvin and Richardson 66 ), whereas four studies (six publications) targeted health professionals( Reference Banz, Most and Banz 67 – Reference Van Der Weijden, Grol and Schouten 72 ) (see Table 1). Of those targeting patients, eighteen measured the primary outcome of cardiovascular risk markers( Reference Arntzenius, Kromhout and Barth 33 – Reference Koertge, Weidner and Elliott-Eller 36 , Reference de Lorgeril, Renaud and Mamelle 39 , Reference de Lorgeril, Salen and Martin 40 , Reference Frost, Brynes and Bovill-Taylor 42 – Reference Goodwin, Forman and Herbert 45 , Reference Masley, Phillips and Copeland 50 , Reference Shenberger, Helgren and Peters 52 – Reference Sundin, Lisspers and Hofman-Bang 54 , Reference Vale, Jelinek and Best 57 , 59 – Reference Van Horn, Dolecek and Grandits 62 , Reference Donner Alves, Correa Souza and Brunetto 64 – Reference Powell, Calvin and Richardson 66 ). Eight of these had diet or nutrition as the only risk factor being targeted( Reference Arntzenius, Kromhout and Barth 33 , Reference Kromhout, Arntzenius and van der Velde 34 , Reference de Lorgeril, Renaud and Mamelle 39 , Reference de Lorgeril, Salen and Martin 40 , Reference Frost, Brynes and Bovill-Taylor 42 , Reference Gleason, Bourdet and Koehn 44 , Reference Masley, Phillips and Copeland 50 , Reference Shenberger, Helgren and Peters 52 , Reference Donner Alves, Correa Souza and Brunetto 64 , Reference Philipson, Ekman and Swedberg 65 ), whereas ten had intervention strategies for multiple risk factors( Reference Billings 35 , Reference Koertge, Weidner and Elliott-Eller 36 , Reference Giannuzzi, Temporelli and Marchioli 43 , Reference Goodwin, Forman and Herbert 45 , Reference Singh, Dubnov and Niaz 53 , Reference Sundin, Lisspers and Hofman-Bang 54 , Reference Vale, Jelinek and Best 57 , 59 – Reference Van Horn, Dolecek and Grandits 62 , Reference Powell, Calvin and Richardson 66 ). All four studies targeting health professionals addressed the primary outcome of changes in practice( Reference Banz, Most and Banz 67 – Reference Van Der Weijden, Grol and Schouten 72 ) and three had diet as their sole focus( Reference Banz, Most and Banz 67 – Reference Perry and McLaren 71 ), with the remaining study targeting multiple risk factors( Reference Van Der Weijden, Grol and Schouten 72 ). In total, seventeen included studies focused only on diet( Reference Aish 30 , Reference Aish and Isenberg 31 , Reference Arntzenius, Kromhout and Barth 33 , Reference Kromhout, Arntzenius and van der Velde 34 , Reference Dalgard, Thuroe and Haastrup 38 – Reference de Lorgeril, Salen and Martin 40 , Reference Frost, Brynes and Bovill-Taylor 42 , Reference Gleason, Bourdet and Koehn 44 , Reference Jackson, Lawton and Jenkinson 47 – Reference Masley, Phillips and Copeland 50 , Reference Shenberger, Helgren and Peters 52 , Reference Timlin, Shores and Reicks 55 , Reference Siero, Broer and Bemelmans 63 – Reference Philipson, Ekman and Swedberg 65 , Reference Banz, Most and Banz 67 – Reference Perry and McLaren 71 ), with the remaining eighteen extending the focus to other risk factors, such as smoking, physical activity or adherence to medication.
MI, myocardial infarction; CABG, coronary artery bypass graft; angina, stable or unstable; CAD, coronary artery disease; PTCA, percutaneous transluminal coronary angioplasty; AMI, acute myocardial infarction; NYHA, New York Heart Association; M, male; F, female; RCT, randomised controlled trial.
All participants in the patient-focused studies were on medications for CVD, with only two studies accounting for the effect of medication by requiring participants to maintain constant dosage( Reference Gleason, Bourdet and Koehn 44 ) or initially stratifying based on dosage of relevant medications, although some dosages for that particular study were changed throughout the study( Reference Philipson, Ekman and Swedberg 65 ). Therefore, all patient studies are reported as measures for the secondary outcome of knowledge or behaviour change relating to dietary intake for consistency. Primary outcome results for the two studies accounting for the effects of medication are included with secondary outcomes in Table 1.
There were twenty-six studies in adults with prior CVD diagnoses (angina: stable or unstable, coronary artery disease, CHD, myocardial infarction, acute myocardial infarction, coronary artery bypass graft or percutaneous transluminal coronary angioplasty)( Reference Aish 30 – Reference Wallner, Watzinger and Lindschinger 60 ). Two studies included populations identified as at high risk of CVD( Reference Gorder, Dolecek and Coleman 61 – Reference Siero, Broer and Bemelmans 63 ), three studies were in those with a heart failure diagnosis( Reference Donner Alves, Correa Souza and Brunetto 64 – Reference Powell, Calvin and Richardson 66 ) and four interventions targeted health professionals treating patients with CVD( Reference Banz, Most and Banz 67 – Reference Van Der Weijden, Grol and Schouten 72 ).
Studies were predominantly randomised controlled trials (n 24, 69 %), with the remainder using uncontrolled pre–post interventions (n 6), pre–post interventions with a control group (n 4; two concurrent and two historical control groups) and one case series. Sample sizes ranged between sixteen and 6428 participants. Studies targeting participants with a prior CVD diagnosis primarily included myocardial infarction (n 10), angina (n 9) or coronary artery bypass graft (n 8). Interventions targeting populations with or at risk of CVD were followed up for between 7 weeks and 6 years, with most following patients for ≤3 months (n 8) or up to 12 or 24 months (n 6 and n 5, respectively). Those targeting health professionals were medium or short term (11 weeks, n 1 and 25 d, n 1 respectively) or unclear (n 2). Population ages ranged from 18 to 85 years and populations were primarily male, with four studies including males exclusively, and ranging from 53 to 90 % male in twenty-two studies. Only two studies focused exclusively on females, with two other studies including more women than men (55 and 96 % female). Studies were implemented primarily in hospital settings (general, teaching and university; n 9), outpatient or cardiac rehabilitation centres (n 7), clinics or general practices (n 5). Three studies used a residential stay as part of their intervention. Five studies( Reference Aish 30 – Reference Kromhout, Arntzenius and van der Velde 34 , Reference Shenberger, Helgren and Peters 52 , Reference van Elderen-van Kemenade, Maes and van den Broek 58 ) were published prior to the release of the first Consolidated Standards of Reporting Trials (CONSORT) statement in 1996( Reference Begg, Cho and Eastwood 73 ) and another ten were published within five years of the release date( Reference Campbell, Ritchie and Thain 37 – Reference de Lorgeril, Salen and Martin 40 , Reference Hofman-Bang, Lisspers and Nordlander 46 , Reference Masley, Phillips and Copeland 50 , Reference Toobert, Glasgow and Nettekoven 56 , Reference Wallner, Watzinger and Lindschinger 60 – Reference Siero, Broer and Bemelmans 63 , Reference Van Der Weijden, Grol and Schouten 72 ).
Table 2 summarises the KT characteristics of the included studies, including the nutrition-focused KT outcomes and associated results. The heterogeneity between KT outcomes and measures precluded any combining of results. Validated dietary intake assessment measurement tools were used in seventeen studies, while twenty-one studies used measurement tools that were unclear or not validated; some studies used a combination of both. Twenty-two studies found statistically significant results for outcomes related to nutrition and/or dietary KT, with eleven being non-significant, mixed significance for outcomes or significance not stated. One study( Reference Dalgard, Thuroe and Haastrup 38 ) found the usual care group (comprehensive counselling) had greater improvements than the brief counselling intervention being tested.
P:S, polyunsaturated:saturated fat ratio; PE, percentage of energy; DINE, unclear abbreviation; BDC, unclear abbreviation; CDC, Centers for Disease Control and Prevention; NCEP, National Cholesterol Education Program; INT, intervention group; CTRL, control group; GI, glycaemic index; sat. fat, saturated fat; CHO, carbohydrate; F&V, fruit and vegetables; AHA, American Heart Association; ↑, increase(d); ↓, decrease(d); FHQ, Food Habits Questionnaire; WFR, weighed food record; 1O, primary outcome; 2O, secondary outcome; ASA-24, unclear abbreviation; NCI, National Cancer Institute; WHI, Women’s Health Initiative; PABA, p-aminobenzoic acid; CALS, unclear abbreviation; ES, effect size; mono. fat, monounsaturated fat; poly. fat, polyunsaturated fat; FRR, food record rating.
1 kcal=4·184 kJ.
* Validated measure: Y, yes; N, no; UC, unclear. ‘Not stated’ was given if no information on the topic could be found and ‘Unclear’ if the description given was too vague to determine which (if any) behaviour change strategy was used.
The theoretical framework for the intervention was not stated (n 18) or was unclear (n 7) in the majority of studies. Ten studies specifically stated the theoretical framework, with no one theory being used predominantly (see Table 2). The most commonly used behaviour change strategies, as defined by Abraham and Michie( Reference Abraham and Michie 27 ), were provision of instruction (e.g. ‘no day without fruit’; n 26), followed by provision of feedback on performance (e.g. from analysis of dietary intake; n 12), the prompting of intention formation (e.g. new resolutions about health habits; n 11) and provision of the behaviour–health link (e.g. information on the influence of diet on blood cholesterol levels; n 8). Five studies reported provision of instruction as the only behaviour change strategy used, with twenty-five studies employing three or more identifiable behaviour change strategies. It must be noted that these behaviour change strategies may not have been used exclusively for diet when other risk factors such as physical activity or smoking cessation were also the target of the intervention.
Many studies reported that the dietary advice provided to participants targeted the reduction in total fat intake (n 12) or asked participants to adopt a Mediterranean or Mediterranean-like eating pattern (n 7; see Table 2). The range of dietary advice provided extended from asking participants to adhere to the particular country’s guidelines for the prevention and treatment of CVD, such as the National Cholesterol Education Program Step 2 Diet, through to general healthy eating. This advice was provided by a range of health professionals (n 9), dietitians (n 6), nurses (n 5), nutritionists (n 2), physicians (n 2), psychologists (n 2) or local experts (n 1, local opinion leader). The nutrition advice provider was unclear in seven studies. All interventions were delivered interpersonally, in either individual (n 16) or group-based sessions (n 13), or a combination of both (n 5), with one study unclear. The degree of intervention standardisation was either unclear or not stated in twenty-five studies; five studies undertook rigorous measures to ensure standardisation and four had varying levels. Further details on the intervention content, resources and intensity are summarised in the online supplementary material, ‘Additional File 4: Intervention content’.
The present review aimed to determine how nutrition-related evidence for the prevention and treatment of CVD is effectively translated into practice. The results indicate that KT is inferred, not stated in this area, and is being under-reported in terms of reproduction for clinicians and policy and decision makers. No studies were identified with the primary aim to translate dietetic knowledge to impact on objective CVD risk markers, which indicates a need for KT strategies in this area to be purposefully conducted and evaluated. The evidence base confirms the relationship between dietary change and improved outcomes of populations living with CVD in clinical interventions, but the KT studies are lacking. Overall, methods describing strategies to initiate and maintain nutrition behaviour changes were of limited value. In addition to this, the measures used to assess the dietary outcomes were varied, with the sensitivity of the tests to determine the extent of change in the outcome of interest unclear in many cases.
Only ten of the thirty-five studies clearly identified a theoretical framework that addressed processes involved in the changing of behaviours. This is in contrast to the fact that all studies required participants to change behaviour if the intervention was to be successfully implemented. The results of larger-scale successful dietary intervention studies such as the GOSPEL study (n 3241)( Reference Giannuzzi, Temporelli and Marchioli 43 ) and the Lyon Diet Heart Study (n 605)( Reference de Lorgeril, Renaud and Mamelle 39 , Reference de Lorgeril, Salen and Martin 40 ) found that dietary behaviour changes were initiated and persisted in the longer term. However, the detail as to how this was achieved was not described and no basis given for why the chosen strategies were appropriate for the context in which they were applied. Therefore replication using the same KT strategy in a given population is not possible and the external validity is unclear. This issue is not limited to nutrition KT strategies in CVD, and has been identified more generally in primary care and general medicine( Reference Glasziou, Meats and Heneghan 74 ). Also, the need for standardised end points for key performance indicators of cardiovascular outcomes in Australia has been indicated( Reference Clark, Redfern and Thompson 75 ). This extends to dietary intake because of the role nutrition plays in the prevention and treatment of CVD. From the present review, it is evident that standardisation of approaches would be of benefit here as well.
In terms of study quality, all interventions focused on risk outcomes important to the prevention and treatment of CVD, although five scored negatively for relevance, as they are unlikely to be replicated in the current financial climate and funding models due to their intensive resourcing( Reference Gleason, Bourdet and Koehn 44 , Reference Hofman-Bang, Lisspers and Nordlander 46 , Reference Sundin, Lisspers and Hofman-Bang 54 , Reference Toobert, Glasgow and Nettekoven 56 , 59 ). Inconsistencies in the published details required to confirm validity may also be due to their older publication dates, as fifteen were published either before or within five years of the first CONSORT guidelines. Of these fifteen studies, only two scored positive for validity( Reference de Lorgeril, Renaud and Mamelle 39 , Reference de Lorgeril, Salen and Martin 40 , Reference Toobert, Glasgow and Nettekoven 56 ). No studies scored positive for all three areas of quality. This may be a publication limitation, as intervention methods were either limited or focused on describing the measures used. Glasziou et al. showed that authors of publications of non-pharmacological interventions, selected for high validity and relevance, were able to provide information supplemental to that published upon request, to provide a more complete description of interventions to aid replication by clinicians( Reference Glasziou, Meats and Heneghan 74 ).
Of the three studies scoring positive quality for KT, only Luszczynska et al.( Reference Luszczynska, Scholz and Sutton 49 ) targeted diet as single risk factor. Luszczynska et al. used implementation intentions training in patients as an adjunct to Phase 2 cardiac rehabilitation. The intervention itself was brief (10–20 min), yet results showed significant changes that persisted up to 6 months. The authors also provided a sufficiently detailed structure of the intervention that could be replicated within a cardiac rehabilitation setting by other health-care professionals. Allen( Reference Allen 32 ) and Goodwin et al. ( Reference Goodwin, Forman and Herbert 45 ) both targeted multiple risk factors, with Allen using Social Cognitive Theory to improve self-efficacy and Goodwin et al. using Acceptance-Based Behaviour Therapy to facilitate participant changes to healthier behaviours. Both studies provide details that are less prescriptive than those found in Luszczynska et al. and therefore more challenging for clinicians to replicate. Allen targeted self-efficacy by the development of specific strategies to attain goals (see online supplementary material, ‘Additional File 4: Intervention content’) and reported positive results after 1 year, but the dietary measure used may not have been of sufficient sensitivity to detect the change in outcome reported of total and saturated fat. Goodwin et al. had a small sample size (n 16) and it is therefore more difficult to determine whether the intervention can be applied to a more diverse population sample, such as those found in current clinical CVD prevention and treatment settings, and attain similar results.
Approximately half of the studies herein focused on diet alone, with the remainder targeting other CVD risk factors as well. The KT results are inconclusive with regard to which is the more successful approach. It has been suggested that it may be easier to translate efficacious dietary patterns, such as the Mediterranean-style diet, instead of focusing on single nutrients, and thereby contribute to better CVD outcomes( Reference Anand, Hawkes and de Souza 76 ). The American Heart Association Scientific Statement (2010) also considered the advantages of focusing on single lifestyle factors compared with multiple factors, such as physical activity, smoking and dietary modification on CVD biochemical risk factors, with a similar inconclusive result( Reference Artinian, Fletcher and Mozaffarian 77 ). However, European guidelines for prevention of CVD in clinical practice recommend multimodal behavioural interventions for individuals at very high risk( Reference Perk, De Backer and Gohlke 7 ).
These constraints may be the result of a difference in focus between explanatory and pragmatic designs in relation to KT. Bhattacharyya et al. highlighted the differences in intervention design between controlled trials that investigate the efficacy of a treatment and pragmatic studies that aim to assess the effectiveness of the treatment in the context of clinical practice( Reference Bhattacharyya, Estey and Zwarenstein 78 ). In particular, the focus of the outcomes from these two very different types of studies necessarily varies greatly. Explanatory designs use process measures for outcomes whereas pragmatic designs use outcomes relevant to health-care stakeholders such as the patients, the health services and funding bodies.
Very few studies were found where nutrition evidence specifically for the prevention and treatment of CVD was passed between researchers and clinicians. There is a gap in the literature about how researchers are passing on their findings of what works to clinicians. While this is most likely taking place in settings such as conferences, seminars and workshops, the translation strategies are not being described, evaluated or appearing in publications. One such example of publication was the study by Banz et al. ( Reference Banz, Most and Banz 67 ) but the study quality was poor which may be due to the short report format.
A number of further shortcomings were identified within the review that reduced the usefulness for KT replication. For example, it was identified that many of the measures and power calculations used may have been of insufficient quality to detect the extent of change in dietary patterns due to the KT strategy. In particular, the dietary measures used cast doubt on the significance of the results. While many studies declared their dietary measures to be validated, it was unclear from the methods whether the instrument used in the study was validated for the outcome for which it was used. Food records and 24 h dietary recalls are regularly used for obtaining data on usual dietary intakes, but adequate standardised protocols for data collection need to be described in the methods to ensure the data are collected in sufficient detail to be considered valid. Many of the participants were also on medication for their condition, which makes the extent of change in biochemical risk markers attributable to diet unclear. This is typified in the study by Masley et al. in which a medication campaign driven by a health-care fund occurred prior to randomisation, accounting for a decrease in LDL cholesterol from 3·7 to 3·1 mmol/l( Reference Masley, Phillips and Copeland 50 ). The use of biochemical risk markers is limited unless medication and other factors are controlled for, such as in the case of Gleason et al. and Philipson et al., where medication was held consistent or results stratified by medication use, respectively( Reference Gleason, Bourdet and Koehn 44 , Reference Philipson, Ekman and Swedberg 65 ). In addition to this, the type of dietary advice people were given, such as replacing dietary fat with carbohydrates of unclear nutritional quality, is likely to have confounded long-term CVD outcomes( Reference Fan, Song and Wang 79 ).
Studies published before the recommendations from the CONSORT statement took effect( Reference Begg, Cho and Eastwood 73 ) were of lower methodological quality. This limitation may have been overcome if the inclusion year were changed to a more recent date, such as to approximately 2001. The rationale for this date includes the release of the first updated CONSORT statement( Reference Moher, Schulz and Altman 80 ) with a 5-year lag after the primary statement to allow reporting to become more consistent. As expected, meta-analysis of data was not possible due to heterogeneity in nutrition KT and outcome measures.
Recommendations for future research and practice
There is a clear need for both efficacy and effectiveness KT trials in the area of dietary prevention and treatment of CVD. Studies need to provide a sound basis for choosing particular theoretical frameworks and behaviour change strategies should be adequately detailed to allow for replication. In addition, study outcomes should use valid measures that are appropriate for the KT and behaviour change strategies, and describe links to a clinically useful outcome. Further information is required as to how clinically useful research findings are effectively translated to clinicians and then patients. Health professionals are the conduits to translating best evidence to at-risk persons, but little evidence currently exists to demonstrate efficacy or effectiveness of the translation link between CVD nutrition research and health professionals. A summary of recommended inclusions for KT publications is given in Table 3.
* These recommendations are specific to nutrition and diet translation studies and are to be used in conjunction with the Workgroup for Intervention Development and Evaluation Research (WIDER) recommendations checklist as per Albrecht et al.( Reference Albrecht, Archibald and Arseneau 29 ).
Acknowledgements: The authors would like to acknowledge the collaboration of Neil J. Spratt and the assistance of research librarian Debbie Booth, who assisted with sourcing appropriate key terms and performed the database search. Additional thanks go to Katherine Brain, Alexa Fryer, Li Kheng Chai and Lee Ashton for verifying extracted results. Financial support: J.R. is funded by a Career Development and Future Leader Fellowship co-funded by the National Health and Medical Research Council (NHMRC) and the National Heart Foundation, and is an investigator on NHMRC programme grant ID1052555. L.N. is funded by an NHMRC Early Career Fellowship (APP1036763). This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. Conflict of interest: None. Authorship: T.L.S. contributed to study design and was primarily responsible for screening and reviewing titles, abstracts and full texts for inclusion, data extraction and manuscript preparation. L.N. and J.R. contributed to study design and manuscript preparation. T.L.B., R.C. and C.E.C. contributed to the study design, screening and reviewing of full texts for inclusion and manuscript preparation. All authors approved the final manuscript. Ethics of human subject participation: Not applicable.
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S1368980016001543