Study design and participants
In a previously conducted randomized controlled trial, participants in the usual care group (UCG) and the Transtheoretical Model-intervention group (TM-IG) were post-hoc classified as true or pseudo-respondents. Details of the intervention have been published elsewhere [8, 9].
The intervention, which aimed to reduce fat consumption, was developed as part of a public health promotion service called the Programa Academia da Saúde, which provides users access to regular physical exercise and nutritional counseling. This service is located in Belo Horizonte, which is the sixth most populous city in Brazil, with 2,375,151 inhabitants [10].
This study involved a sample of participants from the public service aged ≥20 years. The eligibility criteria were as follows: (1) no previous participation in any nutritional intervention that addressed fat consumption, (2) regular use of the service, and (3) written informed consent to participate. The Institutional Review Boards of the University (ETIC 103/07) and the Municipality approved the study protocol (087/2007).
A random sample was used to ensure representativeness. The study population comprised 336 participants. The sample size was determined by a normal bilateral Z-test to compare the proportions of two independent groups, with 80% power to detect a 15% difference between groups at a 5% significance level and assuming 53% attrition. Thus, 168 individuals were selected for the first study phase in 2009, which aimed to validate the algorithm proposed by Greene & Rossi [11]. The algorithm validation phase has been previously described in detail [12]. Briefly, during validation, the algorithm for fat intake and three 24-h dietary recalls (24DRs) were administered every other day, including one weekend day. Initially, the individuals were classified according to the algorithm; then, they were reclassified according to mean lipid intake based on the three 24DRs.
In the second study phase involving the TTM-based intervention in 2009, 118 individuals of the original 168 individuals participated (lost to follow-up: declined, insufficient data, or health or family problems). The participants were simple randomized into the UCG and TM-IG using a random number table prepared by a professor with no involvement in the intervention; the researchers were blinded to the allocations. After postgraduate students obtained the participant’s consent, the appropriate numbered sealed envelope was opened. Among the 118 participants, 97 completed the baseline questionnaire and started the intervention. Additional details about the sample calculation and attrition have been reported previously [8].
Measures
Analyzed data included sociodemographic characteristics, self-reported diseases, food consumption, SOC, and measured anthropometric parameters. The questionnaire was previously validated with service users and was determined to be culturally appropriate for the study population [8]. The questionnaire included questions about sociodemographic variables (i.e., sex, age, educational level [years], occupation status, number of household members, and household per capita income), and self-reported non-communicable diseases. Food consumption was assessed using the three 24DRs and questions about eating habits related to fat.
Anthropometric measurements included weight, height, waist circumference, and hip circumference using techniques recommended by the World Health Organization (WHO) [13, 14]. The measurements were obtained in the morning with the individual fasting, barefoot, and wearing lightweight clothes. Periodically trained dietitians collected all data through face-to-face interviews.
Body mass index (BMI) was calculated from weight and height and classified according to the criteria recommended by the WHO [13] for adults and the Nutrition Screening Initiative [15] for elderly participants. The WHO criteria [14] were used to evaluate waist circumference and the waist/hip ratio. Both circumferences were measured three times, and the means were used for analysis.
Food consumption was investigated as follows: (1) the average of three 24DRs administered on alternate days including a weekend day and (2) direct questions related to dietary lipid intake, such as methods of food preparation, including removal of visible fat from red meat and skin from chicken; types of vegetable oil and milk consumed; and frequencies of consumption of fish and high-fat foods ([never/almost never, <2 times/week, 2–3 times/week, 4–5 times/week, or every day]) [12].
Dietwin software was used to analyze food consumption from the 24DRs. The software was supplemented with a food composition table subsidized by the Ministry of Health of Brazil that was constructed from a laboratory analysis of food chemistry composition and complemented by two other tables used in Brazil [16, 17].
The qualitative adequacy of nutrient intake was assessed using the Institute of Medicine Dietary Reference Intakes.
The SOC for fat consumption was classified using the algorithm proposed by Greene & Rossi [11], which has been validated in Brazil [12]. According to this instrument, the SOC is first classified by evaluating an individual’s perception of lipid intake (“I need your honest opinion about your change of fat intake. Do you consistently avoid eating high-fat foods?”). Second, mean lipid intake values from the three 24DRs were evaluated for the individuals classified into the action and maintenance stages based on self-perceived behavior.
Third, individuals whose self-perceived behavior was not confirmed (i.e., failed to meet the behavioral criterion of fat intake ≤30%) were reclassified through a new self-assessment of their intention to change their eating behavior (Questions: “Do you almost always take the skin off your chicken?”; “Do you often eat reduced-fat or low-fat cheese?”; “You often use light, fat-free, or no salad dressing?”; “Do you sometimes eat fruits as vegetables in snacks?”; “Do you often eat bread, rolls, or muffins without butter or margarine?”) (Fig. 1). Interventions tailored to individuals’ readiness to change were subsequently developed.
To explore the subgroup effects of the previously conducted intervention, secondary analyses of the data were performed. The primary variable was created through post-hoc classification of participants into the PM [6] and non-reflective action [7] stages. Participants were classified into the PM stage if they perceived their lipid intake as adequate, i.e., those classified in the action or maintenance stage in the first step of the algorithm but had an elevated fat intake according to the 24DR food consumption assessment in the 2-step algorithm. The non-reflective action stage included individuals who reported having inadequate fat intake, i.e., those classified in the early stages (i.e., precontemplation, contemplation, and preparation) in the first step despite showing adequate fat consumption (i.e., ≤ 30.0%) according to the 24DR assessment in the 2-step algorithm (Fig. 1).
Intervention
The UCG performed the routine activities of the Programa Academia da Saúde, including aerobic and anaerobic exercise (180 min weekly) and open group education regarding nutrition. The group education sessions occurred monthly and involved topics related to health promotion, prevention, and the control of chronic diseases but not fat consumption.
In addition to these routine activities, the TM-IG underwent TTM-based interventions tailored to the subgroups formed from the SOC classification of individuals for fat consumption: pre-action group and action group [9].
Interventions were developed by a multidisciplinary team that included nutritionists and psychologists theoretically aligned with the TTM. The group technique was concordant with the autonomy principle [8].
The nutritional and health needs of the participants and the four pillars of TTM (i.e., stages and processes of change, self-efficacy, and decisional balance) were considered when developing the themes of the interventions. The interventions aimed to increase participants’ confidence (self-efficacy), raise awareness of the benefits of healthy lipid intake, and reduce the barriers to behavioral changes (decisional balance) [8, 9].
In the pre-action group, the use of cognitive processes of change was predominant, whereas the use of behavioral processes was predominant in the action group. The pre-action intervention aimed to facilitate decision-making through the appreciation of participants’ experiences as well as experienced barriers to changing eating habits. The interventions for the action group emphasized culinary workshops, which allowed individuals to experience daily recipes and the degustation of various dishes. Further details of the intervention are described by Siqueira [9].
The intervention comprised 20 workshops over six months: 10 each for the pre-action and action groups. A total of 54 workshops were held. Additional details about the routine activities and TTM intervention have been reported previously [8].
Statistical analysis
Descriptive statistical analysis was performed. The Shapiro-Wilk test was initially applied to assess the normality of quantitative data. Normally and non-normally distributed variables are presented as mean ± standard deviation and median and interquartile range (i.e., first quartile, third quartile), respectively. The χ
2 test, Fisher’s exact test, Student’s t-test for independent samples, and Mann–Whitney U-test were used for intergroup comparisons. Meanwhile, the McNemar test, paired Student’s t-test, and Wilcoxon signed rank test were used for intragroup comparisons. The level of significance for all tests was 5%.
Data were analyzed with SPSS statistical software (version 16.0, SPSS Inc., Chicago, IL, USA). Prior to the analysis, the consistency of the database was examined by checking that the data recorded in the questionnaire and entered into the database were concordant.