Effect of Behaviour Change Communication on Metabolic Syndrome and Its Markers among Ethiopian Adults: Randomized Controlled Trial

Background: Metabolic syndrome is a global public health problem affecting both developing and developed countries with major consequences on human health, social and economic development. In Ethiopia although there is an increase in the prevalence of metabolic Syndrome due to epidemiologic transition, there is no study that evaluated the effect of interventions. This study aimed to assess the effect of nutrition behaviour change communication on metabolic syndrome and its markers. Method: A individually randomized controlled trial was conducted among Ethiopian adults working in Jimma University from mid of September 2015 to December 30, 2015. A total of 224 participants were randomly allocated into intervention (n=112) and controls (n=112) groups. The list of administrative and academic staff involved in the baseline survey was used as a sampling frame. Data on socio-demographic, anthropometric, biochemical and clinical parameters were collected using trained data collectors. Difference in the differences in metabolic syndrome and its components between baseline and end line were compared by the intervention status. Multivariable logistic and linear regression models were used to isolate independent predictors of metabolic syndrome and its components, respectively. Results: Overall, there was significant difference (P<0.001) in the prevalence of metabolic syndrome between intervention (11.6%) and control groups (37.5%) on the end line survey. On multivariable logistic regression analyses, control groups were 8.5 times more likely to have metabolic syndrome compared to intervention groups (AOR=8.53, 95%CI: 3.60, 20.21, P <0.001). There was a significant mean difference in differences in most components of metabolic syndrome and other lipid profiles except HDL (P=0.717) in the intervention group. The mean difference in differences in waist circumference was 6.3 cm (P<0.001), while that of systolic blood pressure (BP) and diastolic BP were 6.1 mmHg (P< 0.001) and 3.6 mmHg (P=0.001), respectively. Likewise the difference of differences between intervention and control groups was 30.7 mg/dl (P<0.001) for T.Cholesterol, 55.5 mg/dl (P<0.001) for triglycerides, 21.9 mg/dl (P=0.015) of metabolic syndrome. For the intervention group the mean difference in differences was 6.1cm (β=6.1, P<0.001) for waist circumference and 4.2 mm Hg (β=4.2, P<0.05) for diastolic blood pressure and 6.5 mmHg (β=6.5, P<0.001) for systolic blood pressure compared with controls. Similarly, the mean difference in differences was higher in the intervention group by 19.9 mg/dl (β=19.9, P<0.05) for FBS, 57.5 mg/dl for TG (β=57.5, P<0.05), 24.40 mg/dl for LDL (β=24.4, P<0.05) and 30.9mg/dl for T.Cholestrol (β=30.9, P<0.001). This trial is retrospectively registered on Pan African Clinical Trial Registration with unique identification number of PACTR202003465339638. Conclusion: There was strong positive effect of behaviour change communication on metabolic syndrome and its components. The results imply the need for enhancing behaviour change interventions using various strategies at the community and health facility levels to curb the emerging burden of chronic non-communicable diseases in Ethiopia. Future research should examine the sustainability of such behaviour changes using a community based study. were compared by the intervention status. The results showed that there was a significant difference in differences in most components of metabolic syndrome and other lipid profiles except HDL (P = 0.717). The mean difference of differences in waist circumference was 6.3 cm (P < 0.001), while that of systolic blood pressure (BP) and diastolic BP were 6.0 mm Hg (P < 0.001) and 3.6 mm Hg (P = 0.001), respectively. Likewise the difference of differences between intervention and control groups was 30.7(P < 0.001) for T.Cholesterol, 55.5(P < 0.001) for triglycerides, 21.9(P = 0.015) for LDL and 22.2 (P < 0.001) for fasting blood sugar.


Introduction
Since few decades back, the burden of metabolic syndrome and chronic non-communicable diseases is emerging in developing countries at an alarming rate (1)(2). Recent evidence shows that consequent to the globalization, change in dietary pattern and decline in physical activity levels, the magnitude of overnutrition is up surging on top of high prevalence of undernutrition leading to a double burden in sub-Saharan Africa (3)(4).
The global prevalence of non-communicable diseases (NCDs) is on the rise, with the majority of the increase occurring among populations in developing countries (5). In Sub-Saharan Africa (SSA) where Ethiopia is situated although infectious diseases still cause the majority of mortality (69% of deaths), chronic non-communicable diseases such as cardiovascular disease, diabetes mellitus (DM), chronic respiratory disease and cancers, contribute around a quarter of deaths (6). This picture is changing as SSA is undergoing an epidemiological transition with a rapidly increasing burden chronic noncommunicable diseases and associated mortality. In SSA, NCDs are projected to surpass infectious diseases (which are typically endemic to developing countries) by 2030 (5). As the prevalence of chronic non-communicable diseases is increasing, the interface between CNCDs and infectious diseases is becoming clearly apparent implying the need for health intervention strategies that can address both problems. For instance, an increasing prevalence of diabetes may hinder efforts for tuberculosis control, increasing the number of susceptible individuals in populations where tuberculosis is endemic, and making successful treatment harder (3)(4). On the other hand, the high prevalence of HIV infection in developing countries, especially in sub-Saharan Africa and concomitant anti-retroviral therapy leads to an upsurge of metabolic syndrome (7)(8).
Evidence shows that Ethiopia is having burden of metabolic syndrome and mortality from chronic non communicable diseases that are linked to over nutrition and life styles changes. Prevalence of MetS was reported to be 14.0% in men and 24.0% in women using IDF criteria; while individual components are prevalent among an apparently healthy working population in Ethiopia (9)(10).
These findings indicate the need for evidence-based health promotion and disease prevention programs. Several studies in different countries demonstrated that varying levels of life style change interventions can reduce/reverse metabolic syndrome (11)(12). Another study showed that delivery of a structured group education programme to individuals with MetS improved management of cardiovascular and diabetes risk factors (11). However, these issues were not documented in Ethiopian in particular and in Sub-Saharan Africa in general.
Although the magnitude of metabolic syndrome and chronic non communicable diseases is increasing in Ethiopia, there is no any organized public health intervention to prevent the occurrence of these problems and their risk factors. The health care system has limited itself to treating full-blown cases of metabolic syndrome and its components. There no intervention that is underway to address this emerging problem. Studies elsewhere, had demonstrated that dietary intervention can prevent metabolic syndrome (13)(14). The effectiveness of interventions on metabolic syndrome has not been evaluated in Ethiopian set up. The aim of this study was to evaluate the effect of nutrition and life style behaviour change communication (BCC) intervention on metabolic syndrome and its markers.

Methods And Materials Study area and Period
The study participants were recruited from Jimma University, which is a public higher educational institution established in December 1999. Jimma Town is 357 km southwest of Addis Ababa. The university has two institutes and six colleges encompassing a total of 1341academic and 5444 administrative staffs. Circumstantial evidences show components of metabolic syndrome such as diabetes mellitus, hypertension and cardiovascular problems are common even within the small circle of academic and administrative staff leading to morbidity and mortality tolls.
Trial design: An individually randomized controlled parallel design used to determine the effect of nutrition and life style behavior change communication (BCC) on metabolic syndrome and its components.

Participants
The baseline study recruited 704 workers of Jimma University, which was used for developing anthropometric cut-offs for detecting obesity and metabolic syndrome among Ethiopian adults (15).
All administrative and academic staff of Jimma University who was actively working in the university and not away from more than one week during recruitment period was considered for inclusion in the study. Administrative and academic staff of Jimma University who have physical disability including deformity (Kyphosis, Scoliosis), pregnant women, limb deformity that prevents standing erect were excluded from inclusion into the study. The eligibility criteria include: not having any physical disability that interferes with anthropometry, being employ of Jimma University and having baseline data. Those who were not available for the total duration of the follow up were excluded.
Intervention: After giving a gap of 6 months to avoid the periods of feasting and fasting, the BCC intervention continued for three months starting from half of September, 2015 and up to December 30, 2015. The end line data were collected in the first half of September 2015 from the controls and from December 30 to January 15, 2016 for the intervention group to avoid information contamination.
The intervention group was given behaviour change communication through different strategies including power point assisted sensitization, facilitated group discussion using real examples of cases with metabolic syndrome using evidences and effective dietary and life style behaviours. Each of the intervention group members was also given a two page Amharic language brochure to take home.
The study subjects were also reminded about the behaviours using key mottos of the behaviour (eg. Your menu should be colourful, avoid the three whites including: fat, sugar and salt, having diversified diet, aerobic exercise, drink plenty of water, avoid sources trans fats, avoid smoking, avoid khat chewing, avoid excess intake of alcohol, have a quality and adequate sleep and avoid sitting for long time) that were emphasized during training through text messages, while the control group received nothing before data collection (Table 1). After the end line data were collected, similar information was given to the control group.

Randomization
Simple randomization was used to allocate the study participants into the intervention group (exposed to behaviour change communication) and the control group (group which did not receive any intervention). A baseline data were collected from a randomly selected 704 workers comprising of different ethnic groups to develop an optimal anthropometric cut-off for detecting obesity and metabolic syndrome (15). A total 230 individuals were randomly selected using their identification number as a sampling frame. Then they were randomized into intervention (n = 115) and control (n = 115) using ENA Smart software.
Outcomes: Primary outcomes of the study are metabolic syndrome and its components including: waist circumference, systolic and diastolic blood pressures, fasting blood sugar, triglyceride levels, high density lipoproteins, total cholesterol and low density lipoproteins.

Measurements
This study was conducted in accordance with the WHO's Stepwise (STEPs) approach for noncommunicable disease (NCD) surveillance (16). This approach is characterized by the use of  (15) For the end line data, a five ml blood specimen was collected with senior medical laboratory technologist, the serum was separated from the whole blood within 30 minutes and the serum was transferred to Nunc tubes and stored in -20 0 C refrigerators. After all the specimens were collected the serum was packed with ice and transported to Mettu Karl Hospital laboratory and the lipid profile tests were done at this hospital laboratory with HumaStar 80 (Germany) using standard operating procedures. Fasting blood sugar was determined using Humastar within two hours of collection in Jimma University specialized hospital (JUSH) at JUCAN project laboratory.

Statistical Analyses
Data were doubly entered into Epidfata 3.1 and exported to SPSS for windows version 20 (Inc, Chicago, IL, USA) for cleaning and analyses. Student's T-tests was used to evaluate differences in mean values for study groups after checking the assumptions for normality. As the study has baseline and end line data, the effectiveness of nutrition behaviour change communication on metabolic syndrome and its markers was evaluated by comparing the mean difference in differences between the baseline and end line in the intervention and in the control groups in metabolic syndrome and its markers. End line measurements were subtracted from baseline measurements for both intervention and control groups and the differences were compared.
Multivariable logistic and linear regression models were used to isolate independent predictors of metabolic syndrome and its components, respectively. For the multivariable linear regression, all assumptions including linearity, normality, homoscedasticity and multicollinearity were checked using Q-Q plots and multicollinearity was checked using variance inflation factor. For the multivariable logistic regression, multicollinearity was checked using standard error < 2.0 and Hosmer Lemeshaw test (P > 0.05) was used to check model fitness. Statistical significance is set at p-values < 0.05.

Operational Definition of Terms
Intervention group -Randomly selected adults who were given nutrition and health life style behavior change communication for three months.
Control group -Randomly selected adults who were not given nutrition and health behavior change Central obesity was defined as waist circumference ≥ 80 for women and ≥ 94 for men.
Metabolic syndrome (MetS) is defined in accordance with the IDF as presence of abdominal obesity (and three or more MetS components described above (19)(20).

Results
Baseline background characteristics of the study participants are presented in Table 2   In Table 3, comparison of levels of metabolic syndrome components and lipid profiles at baseline and end line was made and the differences of the differences between the baseline and end line values were compared by the intervention status. The results showed that there was a significant difference in differences in most components of metabolic syndrome and other lipid profiles except HDL (P = 0.717). The mean difference of differences in waist circumference was 6.3 cm (P < 0.001), while that of systolic blood pressure (BP) and diastolic BP were 6.0 mm Hg (P < 0.001) and 3.6 mm Hg (P = 0.001), respectively. Likewise the difference of differences between intervention and control groups was 30.7(P < 0.001) for T.Cholesterol, 55.5(P < 0.001) for triglycerides, 21.9(P = 0.015) for LDL and 22.2 (P < 0.001) for fasting blood sugar. Further multivariable linear regression analyses showed that after adjusting for many variables, there was a significant difference in difference between intervention and control groups in the components of metabolic syndrome that are dependent on physical measurements. For the intervention group the mean difference in differences was 6.1 cm (β = 6.1, P < 0.001) for waist circumference and 4.2 mm Hg (β = 4.2, P < 0.05) for diastolic blood pressure and 6.5 mmHg (β = 6.5, P < 0.001) for systolic blood pressure compared with controls. The other variable significantly associated with diastolic blood pressure was baseline body fat percent of the study participants. For a unit increase in baseline body fat percent, the mean difference in differences of diastolic blood pressure was lower by 2.0(β = 2.0, P < 0.05) mmHg ( Table 4).
The results of another multivariable linear regression analyses presented in Table 5 showed that after adjusting background variables, the intervention group had high difference of differences in LDL, Triglycerides, T.Cholestrol and fasting blood sugar (FBS). The mean difference in differences was higher in the intervention group by 19.9 mg/dl (P < 0.05) for FBS, 57.5 mg/dl for TG(P < 0.05), 24.40 mg/dl for LDL (P < 0.05) and 30.9 mg/dl for T.Cholestrol (P < 0.001). Table 5 Multivariable linear regression models predicting mean bassline to end line differences of the differences in the components of metabolic syndrome that dependent on Laboratory analyses of blood sample Variables Baseline end line differences of lipid profiles and Fasting blood sugar  The results showed that in both intervention and control groups small proportion of the study participants were free of any of the metabolic syndrome components (13.4% for intervention vs 5.4% for controls). Large proprtion of both intervention and control groups have atleast one or two of the metabolic syndrome components. Howver, there was a significant difference in the proportion of intervention and control groups with three or more metabolic syndrome componets (Fig. 2).
Overall, there was significant difference (P < 0.001) in the prevalence of metabolic syndrome between intervention (11.6%) and control groups (37.5%). This difference between intervention and control groups was consistent across the different sexes (Fig. 3).
The results of multivariable logistic regression showed that variables including: intervention, sex, age base line body fat percent were independent predictors of metabolic syndrome. Control groups were  (Table 6).

Discussion
The study demonstrated a significant positive effect of behaviour change communication implemented for three months on metabolic syndrome and its components, which is consistent with a 12 week yoga-based lifestyle intervention among Indian adults with metabolic syndrome (21) and a 12 week life style education interventional in Thai adults (22). A systematic review of randomized controlled trials in different population groups also showed positive effect of life style modification interventions starting from 10 weeks duration (23). In this study, the likelihood of having metabolic syndrome was 8.5 times higher among the control group after adjusting for background variables indicating the significance of lifestyle modification interventions in the prevention of metabolic syndrome and related non-communicable diseases. This finding is consistent with the report of meta analyses of randomized controlled trials that showed life style modification intervention were effective in resolving metabolic syndrome (MetS) and reducing the severity of related abnormalities including fasting blood glucose, waist circumference, systolic blood pressure (SBP) and diastolic blood pressure (DBP), and triglycerides in subjects with MetS (12,13,24).
The positive results imply the need for adopting a team approach to lifestyle modification programs in the management of MetS (25). It was suggested that multidisciplinary group approach is an effective and economically feasible strategy in the control of metabolic parameters (26). It was also reported that modifying diet together with frequent physical exercise can reduce the triglycerides concentration as well as SBD and DBP (24,27,28). The need for implementing effective lifestyle modifications to prevent MetS and its health consequences has been indicated (24,29).
Sex and age were other variables associated with metabolic syndrome. It was observed that males were 4.7 times more likely to have metabolic syndrome as compared to females, which is consistent with other reports (30). For one year increase in age, the likelihood of metabolic syndrome was higher by 6%, which is similar to a report from elsewhere (31). The relationship between metabolic syndrome components with older age and male sex has also been reported (32).
It was also observed that the intervention group has significantly (P < 0.05) higher dereference in differences between baseline and end line surveys for serum levels of FBs, Triglycerides and T.cholestrol and higher levels of HDL compared to the control group. A similar finding was reported by studies from elsewhere (24; 27, 33). Likewise, the intervention group had a significant difference in differences in waist circumference similar to reports of studies elsewhere (34)(35).
The results showed that intervention group the mean difference in differences was 4.2 mmHg (P < 0.05) for diastolic blood pressure and 6.5 mmHg (P < 0.001) for systolic blood pressure compared with controls. This finding is consistent with another study which showed that the mean diastolic blood pressure (DBP) and triglycerides decreased significantly in the intervention group in both sexes (36).
A positive effect of life style modification on blood pressure has also been reported among Korean adults (37,38).
It has been documented that interventions incorporating dietary, physical activity and other life style changes exerts beneficial effects on the various components of the metabolic syndrome and improves overall survival (13, 24, 27, 28, 29,). However, interventions targeting life style changes have the potential to succeed only if they are executed early to offer strong evidence to substantiating the development of appropriate public policies (14,23).
Baseline body fat percent of the study participants was significantly associated with diastolic blood pressure such that for one year increase in baseline body fat percent the mean difference in differences in diastolic blood pressure was lower by 2.0(P < 0.05) mmHg. This could be related the effect of high body fat percent on the cardiovascular system which minimizes the difference in the blood pressure changes. The effect of body fat percent on increasing blood pressure has been documented by other studies (13, 38, 39,).
Multivariable linear regression analyses showed that the intervention has high difference of differences in LDL, Triglycerides, T.Cholestrol and fasting blood sugar (FBS). A similar positive effect of life style modification was reported on lipid profiles (40)(41) and blood sugar level (42).
It was also observed that the difference of differences was lower by 30 Similarly, participants who had high income and medium income had higher difference in differences for LDL and T.Cholestrol compared with those in lower income. Evidence shows that an increase in income without change in educational status was associated with poor lipid profiles (44). However, the positive association of income with differences in differences of LDL and T.Cholestrol could be due to the fact that income is highly associated with educational status in the study population (university staff), which in turn may affect the level of adherence to the behaviour change interventions (45).  (46) and history of high proportion of early life stunting (47)(48)(49)(50)(51). Early life stunting could generate a huge potential for emergence of epidemics of chronic non-communicable diseases due to organ stunting. It has been reported that early childhood malnutrition including during the fetal period leads to occurrence of chronic non-communicable diseases later in life (52). The modifiable risk factors include dietary transitions to more processed and low fibre high calorie diet and sedentary and motorized way of life (53)(54)(55). The positive effect of the intervention documented in these modifiable risk factors implies the need for enhancing such interventions using different strategies through the involvement of different sectors to curb the upsurge of non-communicable disease and associated consequences in the years to come.
The study employed an individually randomized controlled trial to generate empirical evidence on the effectiveness of intervention approaches. However, to avoid information contamination the end line data were collected from the control group earlier than the intervention group, which is not expected to have an effect in their dietary and other behavioural factors as the data from the intervention group was also collected in the same season.

Conclusions
There was strong positive effect of behaviour change communication on metabolic syndrome and its components. The results imply the need for enhancing behaviour change interventions using various strategies at the community and health facility levels to curb the emerging burden of chronic noncommunicable diseases in Ethiopia. Future research should examine how sustainable such behaviour changes are using a community based study.

Declarations
The study was funded by Jimma University; Institute of Health. The institute did not have a role in the design of the study and collection, analysis and interpretation of data or in writing the manuscript.

Availability of data and materials
The data used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Authors' contributions
MS and TB conceived and planned the study. MS, DL, MST, TY, TB and ET implemented the study. MS and TB did the analyses.MS drafted the manuscript. DL, MST, TY, TB and ET critically reviewed the manuscript. All authors gave input to the manuscript and read and approved the final version.

Ethics approval and consent to participate
The study protocol was ethically approved by the Ethical review board of Jimma University before the start of the study on February 5, 2015 with a reference number of: RPGC/691/07. Adequate explanation was given to each participant and informed consent was obtained before data collection.
To keep anonymity of all data, no personal identifiers were used except unique ID number. So, unlinked anonymous method was used, where the study subjects and their blood samples were matched later during analyses without revealing any personal identifiers. Informed verbal consent was obtained from each study subjects prior to the administration of questionnaire after the purpose of the study was explained to respondents. The study participants were informed that they will have the right to refuse or discontinue participating in the research without any compromise in the service they are getting from the respective facilities. Finally, the study participants were given their laboratory results printed on a slip that has only their unique identifier. This trial is retrospectively registered on Pan African Clinical Trial Registration with unique identification number of PACTR202003465339638.

Consent for publication
This is not applicable as the study does not have individual person's data. Prevalence of metabolic syndrome between intervention and control groups at the end line by sex, Jimma University southwest Ethiopia