Study characteristic
The basic characteristics of included studies are shown in Table 2.
Type of studies
Nine studies met the inclusion criteria (see Fig. 1). Seven of the included studies were RCT’s [33,34,35,36,37,38,39] and 2 non-RCT’s [40, 41].
Type of participants
In all studies, participants had a mean age ranging from 34 to 72 years. Females represented between 45 and 100% of the overall sample. One study [35] did not state the number of males and females who participated. Six of the included studies had apparent healthy participants [33, 34, 37, 39,40,41]. Three of the included studies had participants with a clinical diagnosis [35, 36, 38]. Of the nine studies, one was carried out in Australia [36], four in the USA [33, 37,38,39], one in Canada [40], and three in the Mediterranean (Greece, Italy and Spain) [34, 35, 41].
Type of dietary pattern
All studies included a whole dietary pattern that took into consideration the main food groups: protein, grains/carbohydrates, oil and fats, dairy, fruit/vegetable (n = 9). Two of the nine studies specifically examined the Mediterranean diet [40, 41], and one examined the DASH diet [38].
Type of primary outcome
Outcome measures varied across studies. Two studies used the HEI-2005 to assess overall diet quality and adherence to the recommended diet [34, 35], with higher scores representing better diet quality. One study assessed adherence to the Mediterranean diet with the Mediterranean Diet Adherence Screener (MEDAS) [41], with higher scores representing higher adherence to the Mediterranean diet. One study used the Diet Guidelines Index (DGI) to measure adherence to healthy recommendations over the previous month. A diet score is obtained with a range of 0–150, with higher scores representing higher levels of healthy eating [36]. One study assessed dietary behaviour with a food frequency questionnaire [49] and compliance to USDA Food Pyramid [33, 50]. One study used the AHEI-2010 to assess diet quality [37] with a total score between 0 and 110, with the higher score representing better diet quality. One study [40] assessed the level of adherence to the Mediterranean diet with a Medscore, which was calculated based on the food frequency questionnaire used in the study. Scores ranged from 0 to 44, with higher scores representing higher adherence to the Mediterranean diet. One study captured recommended foods by a 24-h recall questionnaire and compliance with USDA Food Pyramid [39, 50]. Finally, one study [38] used the Willett Food frequency Questionnaire [51] to derive a DASH adherence score, with a potential DASH score of 1–40 over 8 food components. Each component score between 1 and 5, with a higher score representing higher adherence.
Quality of studies
Out of a total score of 28, all 9 included studies scored between 15 and 25 on the Black and Downs quality assessment checklist (see Table 3), with one study scoring 25 which is considered excellent quality [37]. Four studies scored between 19 and 23 which is considered good quality [33, 36, 39, 40], and the remaining four studies scoring between 14 and 18 which is considered fair quality [34, 35, 38, 41]. Overall, the 9 included studies scored high on the first subscale of the checklist (reporting). None of the included studies met the criteria for “external validity” subscale, with two studies scoring zero [34, 35]. The following section is internal validity-bias which studies scored relatively high on this subsection with scores between 4 and 6 out of a possible 7. The following subsection is internal validity-confounding (selection bias), which yielded the most variety of scores, which may be due to having different experimental designs. Only one of the RCTs [36] reported sufficiently on randomised intervention assignment concealment. Lastly, power to detect a significant effect was reported by 4 studies [33, 37, 39, 40].
Impact of intervention on dietary behaviour
Two studies [40, 41] examined the impact of a theory-based intervention on adherence to the Mediterranean diet. Both studies calculated an overall Medscore pre-post intervention, calculated from the Mediterranean Diet Adherence Screener (MEDAS) [41], or a food frequency questionnaire [40]. Both studies reported a significant increase in Medscore post intervention. One study [38] examined the impact of a tailored behavioural intervention (TBI) on adherence to the DASH diet, compared to a non-tailored intervention (NTI) and usual care (UC) group. At 6 months follow-up, TBI had a higher DASH score than UC and NTI. However, for individual components of the DASH diet such as fruit and vegetables, and wholegrains, there was no significant difference between groups on scores at 6-month follow-up. The remaining 6 studies examined individual components of dietary behaviours based on AHEI [37], HEI [34, 35] DGI [36], FFQ [33] and 24 h recall/MyPyramid [39]. From theses 6 studies, one study reported no improvement in dietary behaviour [36]. Only one study reported a significant improvement in fruit [37], vegetable intake [35], carbohydrates/grains [34] and dairy [34]. Two studies reported improvements in protein (fish, poultry, beans, meat, or eggs) [37, 39] and total fats [33, 34].
Extent of theory use
The extent to which theory was used within the selected studies was assessed using the TCS (Table 4) [26]. From the 9 included studies, the mean total TCS score across studies was 11, which is a moderate application of theory. One study [34] showed a weak application of theory, seven studies [33, 35,36,37, 39,40,41] were moderate, and one study showed a strong application of theory [38]. These scores suggest that theory had not been extensively applied to the design, implementation, and evaluation of behaviour change interventions, and/or theory use was reported with insufficient detail. These scores suggest that most studies are not explicitly reporting theory use in sufficient detail and/or fail to rigorously apply theory to intervention design and implementation. The following section describes the use of theory within the selected studies in terms of the 6 categories of coded items of the TCS [26]: (1) mention of theory; (2) targeting of theoretical constructs;(3) using theory to select recipients or tailor interventions; (4),measurement of constructs; (5) testing of mediation effects; (6) and refining theory.
Category 1: mention of theory (items 1–3)
All studies (N = 9) mentioned a theory (item 1, Table 4), with only 6 studies referring to theory as a predictor of behaviour and provided evidence of the association of the theory or theoretical construct and target behaviour. For example, one study using the Health Belief Model [33] stated that the best predictor of nutrition related behaviour change is the benefit-cost ratio, and for a change in nutrition behaviour to occur, the perceived benefits must outweigh the barriers. Out of the 9 studies, 7 were reported to be a single theory (item 3, Table 4) such as HAPA, SDT and TTM, while 2 studies combined theories (HBM and SCT).
Category 2: are relevant constructs targeted (item 5, 7–11)
Eight of the studies used theory or predictors to select/develop intervention techniques (Item 5, Table 4). Regarding linking intervention techniques to theoretical constructs, only 4 studies explicitly linked all intervention techniques to at least one theoretical construct (Item 7, Table 4), with a further 5 studies linking at least one, but not all, intervention techniques to at least one theoretical construct (Item 8, Table 4). Three studies linked a group of techniques to a group of constructs (Item 9, Table 4). Only 4 studies explicitly linked all relevant theoretical constructs to at least one intervention technique (Item 10, Table 4), with a further 4 studies linking at least one, but not all, constructs with at least one technique (Item 11, Table 4). For example, one study [33] used the HBM to develop an educational intervention to improve dietary practices for CVD prevention. However, the intervention focused on perceived benefits and barriers and neglected other key concepts such as susceptibility and severity of illness, health motivation and perceived control. Another study [39] used the SCT model to develop a dietary intervention and focused their intervention techniques on self-regulation techniques, such as self-monitoring and goal setting, neglecting concepts such as outcome expectancy. Therefore, more than half (N = 5) of these studies did not utilise the full predictive power of their chosen theory.
Category 3: is theory to select participants or tailor interventions
None of the included studies used theory to select participants (Item 4, Table 4), and only 1 study tailored intervention techniques to the participants. Therefore, the intervention differed for subgroups of participants that varied for a particular construct at baseline (Item 6, Table 4). This study was based on the TTM, and the intervention delivered to each participant varied depending on their stage of change at baseline.
Category 4: are relevant constructs measured
Seven of the studies reported measuring theoretical constructs pre-post intervention (Item 12, Table 4), and reporting on the validity and reliability of the scales used to measure constructs/predictors (Item 13, Table 4).
Category 5: testing theory
Seven of the studies reported randomisation, two studies were non-RCTs (Item 14, Table 4). Four of the studies interventions changed the target theoretical constructs. For example, one study [33] using the HBM significantly increased perceived benefits of adoption of positive dietary behaviours and increased nutrition knowledge of CVD and cancer. Also, another study [52] reported that HAPA outcomes in the intervention group reported significantly greater frequency of action planning, and action and coping self-efficacy at follow-up (Item 15, Table 4). Seven of the studies discussed the results in relation to theory (Item 16, Table 4) and three provided support for theory (Item 17, Table 4). That is, studies reported that constructs within the theory, significantly mediated the relationship between the intervention and outcomes. For example, one study [53] that used self-determination theory found that eating related self-determined motivation was associated with an increased adherence to the Mediterranean diet.
Category 6: refining theory
Refining of theory, or suggestions for future refinement was not reported by any of the included studies (Item 18, Table 4).
Fidelity of interventions
Of the 9 included studies, two studies included an assessment on all 5 domains [37, 40]. One study included an assessment on only one domain [41]. Two studies included an assessment on two domains [33, 34]. Three studies included an assessment on three domains [35, 36, 38]. One study included an assessment on four domains [39] (see Table 5).
Study design
All studies made an assessment on study design [33,34,35,36,37,38,39,40,41], with information about treatment dose provided in the intervention condition, and two providing information on treatment dose in the comparison group [37, 39]. All studies reported underpinning theory [33,34,35,36,37,38,39,40,41]. No further trained providers were employed to allow for setbacks.
Training providers
Two studies provided information on training providers [37, 40]. These studies provided information on how trainers were trained and standardisation of provider training. Strategies to enhance training providers included, using the same provider throughout the intervention [40], use of certified trainers [40], and train all providers together [37].
Delivery of treatment
Eight of the studies made at least one assessment on the delivery of treatment [33,34,35,36,37,38,39,40], which was assessed through direct observation of the intervention. Making sure that the interventions were delivered, and the appropriate dose given, being the most reported item in this domain.
Various criteria were used to evaluate the treatment delivery. For example, one study [39] used a checklist after each session to measure degree of adherence, and class attendance [34, 39]. In another study, participants reported on the acceptability of the intervention [36], and how the participants rated the overall delivery of the intervention [40]. Other strategies used to assess delivery of treatment were the use of manuals to aid delivery [33, 36,37,38].
Receipt of treatment
Six studies made at least one assessment on the receipt of treatment. Various strategies were used to assess receipt between authors and included ensuring that participants understood the intervention [35,36,37,38,39,40] and providing resources to enable participants to perform the behaviour [39, 40]. Other strategies to assess receipt of treatment included reviewing self-monitoring data [35, 37], and assessing confidence in behavioural skills [36,37,38,39].
Enactment of treatment skills
Observation and practice of skills required within interventions were included in three of the studies. Observation of these skills in daily life were carried out in two of the studies [39, 41]. Other strategies to assess whether treatment was being enacted were daily self-monitoring and tracking devices [37].