The transtheoretical model is effective for weight management: A randomized controlled trial Intervention for Weight Management

Background: Given the current worldwide epidemic of obesity, interventions with higher impact, such as those carried out in the primary health care (PHC) setting, are needed. Considering that, this study aimed to evaluate the effect of intervention performed according to the stages of change of the transtheoretical model (TTM) for weight management. Methods: Randomized controlled trial in Brazilian PHC that offers free physical exercise and nutrition education. The participants were women aged 20 years or older who were obese or overweight in PHC service. The intervention group (IG, n=51) received the same orientation as the comparison group (CG, n==35) plus individual health counselling based on the TTM aiming weight loss, which lasted six months. The outcome measures were anthropometric, food and nutrient profiles. Inflammatory parameters were evaluated in a random subsample. The inter-group and intra-group differences were evaluated and analysis of covariance (ANCOVA) used to assess intervention effectiveness, using intention-to-treat analysis. Results: There was a difference between groups of -1.4 kg (CI95%: -2.5; -0.3) in body weight after the intervention. About 97% of women in the IG reported benefits of intervention and presented positive changes in diet, biochemical markers and anthropometry. The IG showed a better BMI, resistine and blood glucose result when compared to CG during follow-up. Conclusion: The individualized TTM-based intervention combined with usual care was effective strategy in PHC. These results encouragement of interdisciplinary practices; nevertheless, research to identify additional strategies is needed to address barriers to weight maintenance among obesity low-income women.

behavioral change is complex, unfolding in a sequence of stages (TABLE 1). 7,9 The stages of change (SOC) describe the individual's current intention and engagement toward a targeted health-related behavior. 7,9 In the first stage, called pre-contemplation, the individual didn't have intention to change behavior. The second stage is the contemplation, when the individual is aware that a problem exist, but have not yet made commitment to take action. At the preparation stage, people have intention to take action in the next month. At the action stage, people modify their behavior and at the last stage (maintenance) the individual behavior extends for six months. 7,9 In addition to the SOC, the TTM is based on three other dimensions: (i) processes of change; (ii) decisional balance; and (iii) self-efficacy. Processes of change are activities that individuals engage in when they attempt to modify problematic behaviours. The decisional balance is the comparative potential gains and losses to change and the self-efficacy is the degree of confidence individuals have in maintaining their desired behavior change in situations that often trigger relapse. Decisional balance and self-efficacy varies depending on the stage of change. 7,9 Actions based on these dimensions can help to understand the behavioral change and, therefore, are used to assess the patients' motivation and interventions design. 7,10 Although TTM is widely used to treat various health behaviors, there is little evidence of its effectiveness to induce weight loss. 911 Most evidence are produced in high-income countries and show a reduction of weight, body mass index, and percentage of body fat, 7,12 in addition to an increase in physical activity and an improvement in cholesterol levels and glycemia. 5,13 TTM-based weight loss interventions use multiple behaviors that exhibit greater compliance with body weight control, rather than using a single behavioral approach. 8 Among the most evaluated behaviors are portions control, dietary fat, fruit and vegetable intake, and the practice of physical activity.
Considering the positive results already observed to TTM for weight loss and the lack of interventions to weight loss in PHC context, this study aimed to evaluate the effect of an individualized intervention based on the TTM for the weight management of overweight and obesity women in a Brazilian PHC service. The hypothesis of this study is that interventions based on behavioral theories, like TTM, are more effective for weight loss than the typical primary health care.

STUDY DESIGN AND SETTING
This is a randomized controlled trial comparing two groups, the intervention group (IG) and the usual care comparison group (CG). Eligible participants were recruited from "Programa Academia da Saúde" (PAS) between March to August 2012. The public health system in Brazil has invested heavily in primary care. The system is divided into different health services, such as basic health units and the PAS. This program aims to enhance population autonomy through actions that promote health in socially deprived areas. 14 The PAS service offered, at no cost, physical exercise and food and nutrition education actions conducted by health professionals, for entire population.
More than 2,900 Brazilian municipalities have a PAS unit. The participants have vulnerable situation, with low socioeconomic condition and high prevalence of NCDs 15 . The innovative PAS service represents an important initiative to promote health, prevent, and control prevalence of NCD in especially in populations with high social vulnerability. The Program is an important service of the Health Care Networks, including care of NCDs such as obesity, functioning as a gateway to public health system.
All participants from the PAS unit analyzed were included in the study if they met the following inclusion criteria: female (majority of users in PAS, making it possible to obtain sufficient sample for the trial), 15 have participated in service activities in the previous month (which enabled the nutritional evaluation for sample selection), aged 20 years or older, and obese adult (individuals aged 20 to 59 years and body mass index -BMI ≥ 30 kg/m 2 ) 16 or overweight elderly (individuals aged 60 years or over and BMI≥ 27 kg/m 2 ). 17 The exclusion criteria included individuals with cognitive difficulties, that made it impossible to answer the interview, or who were pregnant because they present specific recommendations for weight control.
The sample of this study included all individuals from PAS unit who met the inclusion and exclusion criteria. Of 294 individuals who were screened, 89 were eligible, and 86 consented to participate (FIGURE 1). The subsample (40%) to biochemical profiles analyses was calculated based an average of other experimental designs, and the participants were randomly selected. 17 The considered sample size effect allows 80% of power to achieve 5% weight reduction as reported in the literature 19 and possible attrition of 30%. A posteriori sample calculation was performed, considering a tolerable relative error of 5%, and 95% of confidence interval, obtaining an average sample size estimated at 39 participants.

RANDOMIZATION AND BLINDING
A blinded researcher performed the participant allocation for CG and IG through a random numbers table. To maximize the proportion of participants receiving the intervention and the statistical power available to compare the groups, the participants were randomized for 40% to CG and 60% to IG. Randomization resulted in the following allocation: CG = 35 and IG = 51. We used separate blocks based on the intervention group to allocate participants to blood collection (CG = 15 and IG = 20). Due to the type of intervention, it was impossible to blind the participants and investigator to the intervention. It was not possible to blind users because of the different nature of the interventions: while the usual care participated in collective activities, the IG had individual counseling. Also, one of the strategies used in the actions with the IG was the agreement of the treatment goals and clarification of the proposed objectives, which precluded its blindness. For all analyses, the participants were assigned to their original groups (IG or CG).

Intervention
Women within the CG maintained usual activities in the service, including physical exercise three times weekly and collective food and nutrition actions once per month. The nutrition actions include collective health education and workshops, with an average of 30 minutes. Women in the IG participated in the same activities as the CG, but they also received individual health counseling based on the TTM for weight management, performed once per month for 6 months.
The TTM was used with health education to foster a reflexive, proactive problem-solving and participatory approach and to consider the different aspects of eating behavior and physical activity to promote the autonomous practice of healthy habits. Therefore, the participants were instigated to identify their obstacles to weight loss and constructed (together with the health professional) possible strategies to overcome them.
Prior to the intervention development, participants were classified into SOC using Weight Loss Behavior-Stage of Change Scale regarding four habits related to obesity: portion control, dietary fat, fruit and vegetable intake, and physical activity. Intervention studies with multiple behaviors show greater compliance with body weight control than using a single behavior-based approach 5,7,10,13 . Among the most evaluated behaviors are portion control, dietary fat, fruit and vegetable intake, in addition to physical activity. These behaviors are key markers for weight loss 7 .
To develop specific interventions, participants were classified into two groups for each habit: (i) pre-action, including women classified in the SOC (TABLE 1), pre-contemplation, contemplation, and preparation; and (ii) action, including participants who were classified in the action or maintenance stages. 17 The approaches were different for the pre-action and action groups for each assessed domain. In pre-action, the IG used processes of change predominantly based on cognitive changes to increase motivation and awareness about health challenges. The action (IG) group focused on processes of change related to behavioral changes, including more detailed guidance on nutrition concepts and physical activity. 9 The intervention was conducted to increase decisional balance and self-efficacy.

MEASURES
Data were obtained from face-to-face interviews in two times: (1) after randomization and before the beginning of the intervention; (2) after 6 months of MTT intervention. The questionnaire included socio-demographic, economic (age, income, education, occupation), and health information (diabetes mellitus, arterial hypertension, self-health perception, body satisfaction, attempted to lose weight in the last month, physical exercise), dietary behaviors (number of daily meals, daily per capita sugar), and food intake (two 24-h dietary recalls), as well as measuring anthropometry (weight, height, waist -WC and hip circumferences). In addition, biochemical parameters [glucose, adiponectin, resistin serum levels] were obtained by blood tests. Inflammatory parameters were chosen for their relationship with obesity.
To analyze the food intake, the mean consumption from the two 24-h dietary recalls were calculated. The 24-h recall examined consumption over two distinct and nonconsecutive days, including weekend or holidays. A kit of homemade measures was provided to improve the estimation of the food quantities. Data obtained from 24-h recall were analyzed with Diet Win ® Professional version 2.0, and the caloric and macronutrient intakes were classified according to gender and age recommendations. 19,20 The subsample participants fasted when blood was collated for biochemical measurements.
Blood collection was performed by nurses in the first and last week of the study, always in the morning, with the individual fasting for 12h. For analysis of the biochemical measures, blood was collected and stored in a freezer at -80°C.
The version of the Weight Loss Behavior-Stage of Change Scale algorithm (validated 5 , translated, back-translated, and adapted for use in Brazil) was used to identify the stages of change for weight management. 21 The instrument was also previously tested in PAS users to verify its applicability. The instrument assesses four habits related to weight management: (i) portion control, (ii) dietary fat intake, (iii) fruit and vegetable intake, and (iv) usual physical activity. For each habit evaluated, the participant evaluated their behavior according to statements ranging from 9 to 11 affirmatives . For each statement, the participant indicated the option that best described their behavior among the five available phrases, each referring to SOC, ranging from "I do not do this at least half the time now and I have no plans to do this" (precontemplation) to "I do this at least half the time now and have been doing it regularly for more than 6 months" (maintenance). The food portion control evaluated the amount of food consumed during a meal, and throughout the day, day-to-day eating control, and impulse to eat control into situations such as stress and depression. The dietary fat intake had information regarding consumption of chicken skin, meat with apparent fat, fat milk, fried foods, fast food, biscuits, butter or margarine, and sauces. The fruit and vegetable intake evaluated the portions consumed throughout the day, as well as their consumption in snacks or substitution for other food, such as sweets. The SOC for physical activity focused on evaluating domestic activities and active daily routines.
To identify the participant's SOC for each of the evaluated habits, the option with the highest number of answers within the questions included in the evaluation was verified, and in case of a tie, the least advanced stage of behavior was considered.
At the 3-and 6-month follow-ups, the IG participants were evaluated for health counseling adherence, including identifying perceived barriers, benefits, and social support to make changes. This information was obtained through the participants' self-report who completed the follow-up.
The primary outcome was changes in weight from baseline to 6 months. Secondary outcomes included BMI, WC, energy consumption, glucose, adiponectin, and resistin. Other measures collected at baseline and 6 months were described. All data were collected by healthcare professionals who were properly trained and supervised by the researchers.

STATISTICAL ANALYSIS
The Shapiro-Wilk test was used to evaluate the distributions of the variables. Variables with normal distribution were presented as means and standard deviations (SDs), and asymmetric variables were presented as medians and interquartile ranges (P 25 -P 75 ). The inter-group differences at the baseline were evaluated using the independent sample ttest, chi-square, and Fisher's exact test for socioeconomic variables, health variables and stages of change. The paired Student t-, Wilcoxon, and McNemar tests were used to evaluate intra-group differences for dieting habits, food intake, anthropometry and biochemical parameters over the study period. The test of homogeneity was performed for each of the primary and secondary outcomes. Analysis of covariance (ANCOVA) was used to assess intervention effectiveness to primary and secondary outcomes, adjusted for age, education, and baseline measurements. The adjustment measures were determined considering important factors associated with the outcomes according to a literature review and the groups differences at baseline. The results of the comparisons were corrected using the Bonferroni method.
Intention-to-treat analysis was performed to determine the effectiveness of the intervention and to keep all participants with non-missing baseline outcome measurements. In case of missing values at 6-month (32.5%), imputation was used by replacing the baseline value following the Baseline Observation Carried Forward approach.
Statistical significance was attributed when p-values were less than 0.05. Analyses were performed using SPSS Statistics for Windows, version 17.0. .

Results
Of the 86 eligible women enrolled at the beginning of the study, 58 completed the 6month intervention: 24 women in the CG (31.4% of attrition), and 34 in the IG (33.3% of attrition) (p = 0.85) (FIGURE 1). In the CG, losses were related to non-attendance to the face-to-face interviews three consecutive times (n = 11); these participants were excluded from the study. In the IG, losses were related to non-attendance three consecutive interviews (n = 6), as well as infrequent use of the service and without telephone contact (n = 3), those unable to continue the nutritional intervention because of personal problems (n = 6), and participants who refused to continue the nutritional intervention (n = 2). The CG participants who completed the trial had a lower per capita income than those that did not complete the trail [165.87 (86.67-240.00) and 345.87(180.00-244.00) respectively; p = 0.008]. Completers only analysis found the same results for anthropometry and dietary habits (data not shown).
At baseline, the CG and IG participants were approximately 55-years-old and with a BMI above 30 kg/m 2 (30, 8). There were significant differences in education, self-health perception and physical activity intergroup. Both groups reported low income per capita, low educational attainment, and did not perceive their health as well. Overall, most women were dissatisfied with their body shapes and had previous weight loss attempts.
Most of the participants were classified in the action and maintenance SOC for all four aspects (TABLE 2).
After 3 and 6 months of intervention, approximately 97.0% of women in the IG reported having followed the intervention guidelines. The main perceived barriers to adherence were lack of time, financial difficulties, and lack of desire/motivation. Approximately 20.0% of the participants did not report adherence difficulties (TABLE 3).
Over 90% of participants who completed the 3-and 6-month follow-ups reported benefits associated with the health intervention (TABLE 4), which was confirmed by changes in eating habits and anthropometric and biochemical profiles observed at follow-up (TABLE   5).
Analyses of the intervention effectiveness after adjusting for age, education, and baseline measurements revealed significant differences in body weight, BMI and resistin between women in the IG and CG (TABLE 6). However, no difference was found in the secondary outcomes, included WC, energy consumption and adiponectin.

Discussion
The TTM-based intervention was shown to be an effective strategy for weight reduction in PHC leading to positive effects concerning nutritional status, dietary behaviors, waist circumference, glucose and resistin levels. The usual care within PAS was unsuccessful in controlling weight during the 6 months of the trial. These findings suggest the promise of the TTM-based intervention approach for weight-loss behaviors while indicating the need for more effective weight maintenance strategies for usual public health care and communities of low socioeconomic status.
To our knowledge, this is the first study in the literature that investigates the impact of an individualized multi-behavioral TTM intervention that addressed multiple behaviors for weight management on PHC, associated with inflammatory and metabolic evaluation.
Furthermore, food and nutrient intake, eating habits, anthropometric characteristics, and biochemical and inflammatory profiles were simultaneously assessed.
The use of multi-behavioral interventions is a promising strategy to control multifactorial morbidities (such as overweight and obesity), especially when considering the complexity of behaviors associated with individual lifestyles. 12 Moreover, it should be considered as a maximized cost-effectiveness intervention, promoting action and paired actions, in which a change of one behavior can help to change another behavior. 11 To overcome impersonal programs interventions, this study used an individualized intervention according to readiness for change that considers the complexity involved with food choices. Feasible strategies based on the economic, cultural, and social environments of participants, which enabled them to make healthy choices were developed. At the 3and 6-month follow-ups, the vast majority of participants reported that they followed guidelines. Adherence to treatment is one of the main determinants of its success and is influenced by different factors, such as recognition of risk behaviors and development of self-care 22 . Also, participants perceived different health benefits with TTM-based intervention. This finding makes sense because TTM enables differentiated interventions according to individual subjective perception.
Despite following the guidelines, participants identified barriers to change that were addressed throughout the intervention. After 6 months, the main barrier identified was the motivation to follow weight loss treatment. This demonstrates the need for constant assessment of readiness for change to guide the intervention. In addition to specific guidelines to the SOC, strategies such Social Cognitive Theory and Motivational Interviewing have proven effective for this type of obstacle 14 .
Hence, while the CG presented unfavorable changes, the relatively high levels of adherence in IG were accompanied by beneficial changes in anthropometric measurements, diet habits, and biochemical parameters. The results suggest that the intervention increased their awareness of diet, but this must be confirmed through other studies. This probably encouraged participant autonomy and most likely led to changing parameters beyond those specifically addressed by the intervention. 9,23,24 Another important improvement presented by IG is the progression from early stages of change to final stages (action and maintenance) regarding portion control, which suggests increased food selectivity and impulse control, which in turn is an important indicator of the sustainability of behavior change. 9,23 The SOC evolution and the eating habits changes that occurred in the IG might have helped the weight loss to occur. The differences between the IG and CG are relevant, especially considering the obesogenic environment of large cities and the positive effects that modest weight loss might have on reducing metabolic and cardiovascular risks. 24 Similar results in primary care were observed in a TTM-based weight loss trial with a lowincome sample, showing weight loss after 6 months. 12 Another trial using multi-behavioral TTM intervention also showed significant weight loss. 13 Some clinical trials that used different approaches, such as online and for telephone, reported weight loss ranging from 0.2 to 2.1 kg. 7,11,25 Despite being one of the target behaviours of the intervention it is important to note that participants in both groups were not different in exercise practice throughout the trial (data not shown). The PAS routinely offers regular physical exercise, and all the participants were users of the service prior to the study. Therefore, everyone practiced physical activity, which is valuable for inflammation control, whether resulting from an isolated effect or as a precursor to reduced body weight. 26  to nutritional guidelines was not evaluated for participants who abandoned treatment. The difference between decisional balance and self-efficacy along the trial were not evaluated, but these constructs were considered for the development of the nutritional intervention.
Since women from the CG and IG participated in the same usual activity, comments the contamination inter groups was possible. Another limitation of this study was the difference in educational level between participants in the CG and IG groups. However, all analyses were adjusted by education to minimize the impact of this variable on the results. Significant results of body weight change were not found in the intra-group analysis. However, although not significant, the results showed a tendency to reduce weight in the IG and gain in the CG. The inter-group comparison showed this difference in a significant way.
In 6 months, the attrition of both groups was 32.5%, reflecting the challenge of conducting an intervention study without offering payment to subjects. This might also reflect the PHC service, which presents a high turnover of users. 29 Intention-to-treat techniques were used to minimize the impact of loss.
In addition, the methods of this clinical trial did not allow blinding of the participants and professionals involved. Another important limitation is the possible contamination between CG and IG, which were carried out in the same place.
The data collection of this work was carried out in 2012. Differences in the context experienced today and in the time of data collection should be considered when interpreting the results. However, it is important to note that after the study, PAS was expanded in Brazil and is today considered a useful space for the treatment of obesity and it is part of the network of attention to chronic diseases in the country. 14 Additionally, this study is performed in a PHC service in Brazil; the results should not be generalized to populations with other characteristics. More studies are needed for other populations. This study showed that the TTM-based intervention for weight control in this scenario was effective and shows the relevance of these strategies in PHC for weight control.

Conclusions
The individualized TTM-based intervention combined with usual care might offer a viable and effective strategy for weight loss in primary health care and have a positive impact on inflammatory markers. The characteristics of the approaches used are in line with international guidelines aimed to prevent and control NCD by promoting encouragement of interdisciplinary practices that are indispensable for successful treatment of obesity and other NCDs. Nevertheless, further research to identify additional strategies is needed to address barriers to weight maintenance among obesity low-income women and sustainability assessment of changes.

AVAILABILITY OF DATA AND MATERIALS
The data sets in this study are available from the corresponding author on reasonable request.

COMPETING INTERESTS
The authors declare that they have no competing interests. had no role in the study design, collection, analysis or interpretation of the data, writing the manuscript, or the decision to submit the paper for publication.

AUTHORS' CONTRIBUTIONS
PPF contributed to designing the study, analysing the data, writing the article and have approved the final manuscript.
MCM contributed to designing the study, interpretation the data, critical review of content and have approved the final manuscript.
LCS contributed to designing the study, interpretation the data, critical review of content and have approved the final manuscript.
AMP contributed to designing the study, interpretation the data, critical review of content, have approved the final manuscript AVMF contributed to designing the study, interpretation the data, critical review of content and have approved the final manuscript.
ACSL contributed to designing the study, analysing the data, writing the article, critical review of content and have approved the final manuscript.