Discussion:
The current study tested the utility of an intervention informed by an integrated theoretical model to promote low-fat food consumption, carbohydrate counting, and physical activity among adults with T2D in Iran. This theory-based intervention could be adopted by chronic disease health care professionals. Based on the results of earlier qualitative formative research among the target population (Blinded for review), the intervention focused on constructs from the TPB and risk perceptions and planning factors from the HAPA. To our knowledge, this study comprises the first formal evaluation of a theory-based intervention to promote carbohydrate counting - a critical behaviour in diabetes management. The results suggest that the intervention was successful in changing behaviour and cognitions related to carbohydrate counting, showing some support for the integrated model to inform behaviour change in this context. However, it did not produce the expected changes for either low-fat food consumption or physical activity. The intervention was only successful in promoting the volitional construct of planning, but behaviour and other cognitions remained unchanged. This may be due to the nature of diabetes as a disease highly concerned with nutrition behaviours such that people already associate a diabetes diagnosis with a need to address any unhealthy eating patterns (Yannakoulia, 2006) .
For carbohydrate counting, the intervention showed promise by significant increases in behaviour, intention, PBC a,nd planning for intervention participants over time compared to control participants, with the improvements in carbohydrate counting behaviour at least partially due to enhanced planning. These findings emphasize the importance of including planning as a self-regulatory component in behavioural interventions, reinforcing assertions that the impact of cognitions on behaviour is mediated by planning (Norman and Conner, 2005, Sniehotta et al., 2005a, Gellert et al., 2012). There were no improvements over time among intervention participants for the motivational phase constructs of attitudes, subjective norm, and risk perceptions, but there was a significant change for PBC. This highlights the importance of control perceptions and the need to consider barriers to performance at both motivational and volitional stages of decision making for carbohydrate counting.
The observed differences for carbohydrate counting, as opposed to low-fat food consumption, may be due to the specificity of the carbohydrate counting behaviour for diabetes management and to the short-term acute symptoms resulting from non-compliance (leading to abnormal blood glucose and, consequently, clinical outcomes). Further, participants had completed the study inclusion requirement of diabetes education, incorporating carbohydrate counting training. They so were at least primed as to the importance and positive benefits of this crucial diabetes management behaviour.
For physical activity, intervention participants showed an improvement in their degree of planning for physical activity compared to control groups but no other cognitions or activity behaviours. These findings are in contrast to other TPB-related studies showing it to be an effective basis for a physical activity behaviour change intervention among adults with T2D (White et al., 2012). The current study may have faced challenges in promoting physical activity due to: (a) most participants were females who are culturally less involved in physical activities, especially outdoor exercise, and (b) >60% of participants had full-time jobs. The commute in Tehran is often 3-4 hours/day, both barriers which may benefit from direct targeting in future interventions.
The intervention significantly reduced TG levels by approximately 10 mg/dl but had no effects on LDL-c, weight, and BMI. Low-carbohydrate diets have been shown to reduce TG rather than LDL-c (Mahan et al., 2012, Santos et al., 2012). As the only behaviour change observed in this study was for carbohydrate counting. Because carbohydrate counting restricts carbohydrate intake explicitly, it seems reasonable to relate the reduction of TG to carbohydrate counting behaviour. Congruently, a meta-analysis by Santos et al., showed that low-carbohydrate diets did not affect LDL-c (Santos et al., 2012).