Central obesity, not BMI explains cardio-metabolic risks among university employees, Ethiopia – a cross-sectional study

Background: Evidence suggests that the middle and low-income countries such as Ethiopia are facing the growing epidemic of both communicable and non-communicable diseases creating burden on the healthcare system. The increasing prevalence is attributed to sedentarism, lifestyle changes and the presence of other cardio-metabolic risk factors. Therefore this study was designed to assess the prevalence and association of overweight, obesity, and cardio metabolic risks and to explore if there was any agreement among the anthropometric measurements among the academic employees of the University of Gondar, Ethiopia. Methods: An institutional based cross-sectional study was conducted using the WHO stepwise approach and recommendations on 381 academic staff of the university. In addition, physical measurements such as weight, height, waist and hip circumferences, and biochemical measures such as blood pressure and fasting blood glucose level (peripheral blood samples by finger puncture) were measured using standardized tools. Result: The mean age of the participants was 33.5 years. The prevalence of obesity among the study participants defined by body mass index, waist circumference, waist height ratio and waist hip ratio was 13.1%, 33.6%, 51.9% and 58.5% respectively. The prevalence of diabetes was 4.7% among which 1.3% was not diagnosed prior to this study. About 53 (13.9%) of the study sample were found to be hypertensive (6.3% known versus 29 7.6% newly diagnosed). Among the participants 39.4% and 23.4% were found to be pre hypertensive and pre diabetic respectively. WC was significantly associated with HTN (AOR = 5.14; 2.50, 9.72), pre-DM (AOR = 4.03; 2.974, 5.96), DM (AOR = 3.29; 1.09, 6.01). In addition, WHtR was significantly associated with Pre-HTN (AOR = 2.69; 1.49, 4.58), HTN (AOR = 2.06; 1.01, 6.31), and DM (AOR = 1.85; .76, 4.32). Conclusion: This study results revealed

among the participants. The result of this study suggests that the constructs of central obesity, not BMI has to be used to screen risks of cardio-metabolic risks among Ethiopians. Keywords: Diabetes mellitus, Hypertension, pre-hypertension, pre-diabetes, Ethiopia Background The increase in the global prevalence of non-communicable diseases in the past four decades in high-income countries is well documented (1)(2)(3). There is evidence to suggest that low-income countries face the triple burden of undernourishment, communicable and non-communicable diseases attributed to low socio-economic status, lifestyle and nutrition changes (4)(5)(6). In sub-Saharan Africa (SSA), about 69% of mortality is caused by infectious diseases, while non-communicable diseases (NCDs) contribute around one-fourth of the deaths (7,8). However, due to the ongoing epidemiological transition in these countries, there is a projection that death due to non-communicable diseases is estimated to surpass death from infectious diseases in the year 2030 (9,10).
Ethiopia is among the countries in the sub-Saharan region that is facing an increasing burden related to non-communicable diseases. For instance, in previous Ethiopian studies, the prevalence of metabolic syndrome (Mets) and diabetes among the working population was reported as 14% and 6.5% respectively, while the prevalence of high blood pressure among these population ranges from 19.6%-30.3% (11)(12)(13). The international diabetic federation (IDF) in its 2014 report estimated the number of people living with diabetes to be 4.9 million and about 2.9 million people living with impaired blood glucose in Ethiopia.
In addition, the report also estimated that the prevalence of undiagnosed diabetes to be about 1.4 million being higher in urban than rural population (14,15). Moreover, among urban dwellers in Ethiopia, about 64.8% of hypertensive subjects and 53.4% of diabetic subjects were estimated to be undiagnosed (16,17).
Previous regional studies used IDF criteria to estimate cardio metabolic risks and association of anthropometric indices among bank workers, school teachers, policemen, and other general population (18)(19)(20). The IDF recommends Asian and SSA to use the driven cut off point to define obesity which is a known indicator and predictor for cardio metabolic risks (20). As a result, many studies in Ethiopia used the recommended cut off point to define obesity. Though it is recommended to use more reliable population specific anthropometric cut off measures and it is well known that slender body framed Ethiopian and some Asian countries still use international or European anthropometric cut off values to identify cardio-metabolic risks (21,22). Hence, it is a high time for Ethiopia to initiate small and/ or large scale high risk approaches for screening NCDs in susceptible populations using simple and sensitive screening tools. Anthropometric indices are simple and effective indicators of general and central adiposity in identifying cardio-metabolic risks in the community and institutional populations in low-income countries (23,24).
Evidence substantiates the presence of correlation between sedentism and the increased chances of cardio-metabolic risks among different population (25-27). Since University academic employees are more likely at risk of sedentism, occupational status, income, and other lifestyle changes, it is crucial to study cardio-metabolic risks and its associated factors among this population.. In Ethiopia, population-based data on the prevalence of diabetes; IFG and hypertension are scant unlike elsewhere, where these data across various occupational groups is recent interest (27-29). Current knowledge of cardiometabolic risks in Ethiopia is mostly based on sporadic, convenience based, and hospital data from patients who present with NCDs. Considering the devastating effects of cardiometabolic health problems on individuals living with chronic illnesses and the country at large, there is an urgent need and impartial attention of exploring cardio-metabolic health problems and its underlying risk factors in Ethiopia. Therefore, the aim of this study was to assess the prevalence of general obesity and central obesity and the association of these anthropometric indices with cardio-metabolic risks such as hypertension, prehypertension, impaired fasting glucose, and diabetes among academic employees in Northwest Ethiopia.

Study design and subjects
This institutional based cross-sectional study was conducted on 381 academic staff working at the University of Gondar, Gondar, Ethiopia. The University of Gondar is one of the oldest and highest educational institutions in Ethiopia located at 750 km north to capital city Addis Ababa. A total of 1387 Ethiopian academic staffs (1221 male, 166 female) who were working as permanent staff were considered as the study population.
The study was conducted from November 2017 to January 2018.
The sample size was determined using Epi info version 7.0 (Centers for Disease Control and Prevention, USA). A two population proportion formula was used to calculate the desired sample size based on the previous study (25) with obesity in hypertensive and obesity in normotensive subjects as 45% and 30% respectively with an odds ratio of 1.91.
A power of 80% to detect real association of exposure variable and 95% level of confidence was used. The derived sample size was n = 390, accounting for estimated refusal or non-response rate of 10 % the required sample size was n = 429. Obesity with DM and hypertension was taken from various regional studies. Finally, hypertension with the highest sample size was chosen.
Among the 1387 permanent academic staffs list secured from UOG, human resource department 429 questionnaires were distributed proportionately and randomly selected staffs in this survey. Socio-demographic, anthropometric indices, and cardio-metabolic risk factor data of recruited staffs were used in this study. About 62 Expatriate staffs, 5 subjects with abdominal surgeries within previous 3 months, 9 pregnant  The cardio-metabolic risk was defined as; Pre-diabetes (Pre-DM) was diagnosed if the FBG value was ≥100 mg/dL to ≤120 mg/dL. Diabetes was diagnosed if FBG value was >120 mg/dl, self-reported Type II diabetes mellitus, or diabetes medication. Pre-hypertension (Pre-HTN) was diagnosed if systolic blood pressure (SBP) >120 to <140 mmHg and/or diastolic blood pressure >80 to <90 mmHg. Hypertension (HTN) was defined as SBP ≥ 140 mmHg and/or DBP ≥ 90 mm Hg. analysis. Mean and percentage with 95% CI adjusted for age was used to describe the distributions of socio-demographic characteristics, anthropometric measures and cardio-metabolic of the study population by cross-tabulation and independent t-test across gender. Pearson's chi-square was used to evaluate the difference between a categorical variable and age-adjusted one-way ANOVA was used to compare continuous variables.
Pearson's correlation test was done to measure the association between anthropometric measures (BMI, WC, WHtR & WHR), age, and cardio-metabolic risk variables (FBG, SBP, DBP). Age controlled bivariate analyses were conducted with the dependent variables (FBG, SBP, DBP) and independent variables that were found statistically significant were included in multivariate analysis. When clear subgroups seemed present in the data, significance testing (Pearson χ 2 ) and, if appropriately sized subgroups per category remained, logistic regression was performed. The prevalence estimates for obesity defined by BMI, WC, WHtR, and WHR was determined separately. In all cases p-value < 0.05 at 95% confidence interval was considered statistically significant.

Results
A total of 429 academic employees were approached for consent, out of which 381 employees (330 male, 51 females) consented and completed questionnaire, physical measurements, and biochemical measurements. The response rate was 89% and this is 97.6% of the power calculated sample size (n = 390). The mean age (in years), height (cm), and weight (kg) of the total participants were 34.33, 164.15 and 64.8 respectively with significant difference between gender. Majority of the participants (64.8%) were below the age of 35 years and 339 (88.97%) participants of the total sample were less than 45 years old.About73% had postgraduate education level and no significant difference in educational level between genders. Among the participants, 6.3% and 7.6% were known HTN and newly diagnosed HTN, while 2.9% and 1.6% were known DM and newly diagnosed DM respectively. About 15 (3.9%) participants had concurrent HTN and DM (Table 1). A gender wise statistically significant difference was observed between age, height, weight, and all anthropometric measures. Majority of the participants (n = 239, 62.7%) were diagnosed either or both prehypertension and pre-diabetes.
The age-adjusted mean of overall BMI was 23.4 kg/m 2 and 24.8 kg/m 2 for men and women respectively. Age-standardized prevalence of pre-obesity in men (23.6%), and women (23.5%) and the prevalence of obesity are; men (3.3%) and women (9.8%). Mean BMI and the prevalence of pre-obesity or overweight and obesity by age group are shown in Table   2.
Gender wise occurrence and mean of cardio-metabolic risks in both general obesity by BMI and central obesity by WC, WHtR, and WHR are shown in Table 4. Higher rate of pre-HTN and HTN in women was observed in central obesity groups whereas in men both general and central obesity groups showed close to similar occurrence. Higher rate of pre-DM was observed in central obesity group defined by WC in both men and women 58.3% and 64% respectively.
Among men, a significant positive correlation was seen between BMI, WC, WHtR, and WHR with SBP, DBP, and FBG. Among women, BMI showed a poor correlation with all cardiometabolic risks. Pearson's correlation for the association between cardio-metabolic factors and anthropometric measure and age for both genders are shown in Table 5. In addition, all central obesity measure showed a significant positive correlation with cardio-metabolic risks except for WHtR and WHR with DBP which were non-significant.
The overall age-adjusted odds ratio of pre-HTN, HTN, pre-DM and DM for general obesity by BMI and central obesity defined by WC, WHtR, and WHR are shown in Table 6. WC was significantly associated and was the most sensitive for HTN (OR 5.14 p < 0.001), pre-DM (OR 4.03 p < 0.001) and diabetes (OR 3.29 p < 0.02). WHtR was significantly associated with pre-HTN (OR 2.690), HTN (OR 2.066) and diabetes (OR 1.855). However, BMI and WHR were found not to be significantly associated with the cardio-metabolic risks included in the multivariate and bivariate regression model respectively.

Discussion
This study is unique in investigating the association between anthropometric measures which includes central and general obesity and cardio-metabolic risk factors such as Diabetes Mellitus and hypertension among the university academic employees in Ethiopia.
This study results revealed variable prevalence between general obesity (27.8%) and the anthropometric indices (IDF cutoff) defining central obesity; WC, WHtR and WHR was 33.6%, 52%, and 58.5% respectively among the participants. In the current study more than half of all adults with excess body fat defined by WC, WHtR, and WHR were defined as non-obese according to BMI, which shows the possible bias associated with the use of BMI and underestimates of obesity and higher sensitivity of measures of central obesity in the study population.
With a higher prevalence of central obesity among the study participants and its association with pre-diabetes, diabetes, pre-hypertension, and hypertension explains that these comorbid conditions are reaching epidemic proportion in Ethiopia. Moreover, it is surprising to observe newly diagnosed diabetes (1.3%) and hypertension (7.6%) among the study samples given these population attained higher literacy level. This study also found that measures of central obesity defined by WC, and WHtR but not general obesity were strongly associated with cardio-metabolic risks in men and women. Though, the prevalence of WC was lower among measures of central obesity it showed the strongest association with HTN, pre-DM and DM while WHtR showed association with pre-HTN, HTN, and DM in our study participants. This is very unlike the findings reported elsewhere, while the difference could be explained by the uniqueness of the stature of Ethiopians (32).
Our study revealed that there is a high prevalence of pre-obesity among male (29.4%) and female (24.8%) using BMI measurement. To the contrary, the prevalence of obesity among men and women was 5.4% and 11.5% respectively. This is rather an intriguing finding and reasons might be that the majority of the study participants were younger than 35 years old and may eventually become obese if not intervened.
The overall prevalence of overweight in the current study was 27.1%, which is much higher than a pooled prevalence of overweight of 15.9% reported in Demographic and Health Survey (DHS) of 32 Sub-Saharan African countries, ranging from 5.6% in Madagascar and 27.7% in Swaziland (33). However, the prevalence of obesity defined by BMI in this study is within the range reported by the DHS study. According to 2016 WHO report, the estimated prevalence of overweight or pre-obesity was 39% among men and 40% among women, while the estimated prevalence of obesity was 11% among men and 15% among women, which is much lower than the figures reported in this study (34).
Regardless, the trending of the prevalence of obesity in Ethiopia as reported by previous study (35), including this study requires attention. More interestingly, the obesity defined based on WC, WHtR and WHR were highly prevalent than defined by BMI. And it is worth to note that slender body framed Ethiopians have higher body fat at relatively low body mass index compared to study samples from other countries (32). Systematic reviews of a large amount of high-quality and consistent evidence show that the use of BMI to define obesity (the degree of excess body fat) might be highly specific, but has low to moderate sensitivity when compared with obesity defined by WC, WHtR, and WHR (36,37).
The overall prevalence of hypertension and diabetes among the participants were 13.9% and 4.5% respectively. Among which a large proportion of them were undiagnosed and thus untreated. In addition, this study also found that 39.4% of the participants were prehypertensive and based on IDF criteria 23.4% were pre-diabetic which implies that these people are at risk of developing HTN and diabetes in the future if not intervened early.
The prevalence of hypertension reported in this study involving healthy population is significantly higher than that of earlier reports in Ethiopia which was 7.1% and 1.8% (38,39) of two and three decades ago respectively which is suggestive of the ongoing salient rise in this country. The reasons for higher undiagnosed cardio-metabolic risks might be due to low health seeking behavior among Ethiopians and if this is the case among the educated study population, it might be a major concern among the larger illiterate Ethiopian population.
However, the prevalence of hypertension reported in this study is lower than other community-based studies in Ethiopia which is 28.3% (40) and 30% (41). The prevalence of diabetes in the current study is consistent with the growing body of regional evidence particularly urban dwellers (42,43). Unlike many other regional studies (40, [43][44][45], the odds of likely cardio-metabolic risks in this current study population is only explained by measures of central obesity and not by body mass index. Moreover, surprisingly WHtR which is proposed (46) as a good indicator of abdominal obesity (AO) and a better predictor of cardio-metabolic risk had lesser odds of likely cardio-metabolic risks than WC in our study participants. This might be due to over estimation because WC does not take height (risk of tall) into consideration while defining AO and unique morphology of Ethiopians. About 91% of the current study samples were aged between 25 and 45 years and hence, higher prevalence of prehypertension and impaired fasting glucose risks among them based on central obesity (WC and WHtR) as indicated in this study should stand as an alert for Ethiopia.
Although there is variability of the prevalence of pre-diabetes in different literature, our finding suggests that 23.4 % of our study samples are pre-diabetic. This figure is slightly higher than a cross-sectional study conducted in Kenya 18% and 8.6% reported by a population study conducted in Uganda (47,48). Likewise, our finding indicated a significantly higher prevalence of pre-hypertension among our samples when compared to a report from Iran 33.7% (49), while it is lower than what is reported in a Nigerian study 45.5% (50). This variability in the prevalence of pre-hypertension and pre-diabetes among different literature could be attributed to the population characteristics in the studies.
Regardless, the finding from our study suggests that the necessity of designing health promotional activities which promote early health screening of cardio-metabolic risks such as diabetes and hypertension. In addition, since the majority of patients with diabetes and hypertension would undergo a long time of pre-diabetes and pre-hypertensive duration, it is essential to design target interventions to either reverse or slow down the progression of these conditions before it becomes full-blown diabetes or hypertension.
This study has reported a well-powered insight into the prevalence of self-reported and measured hypertension and diabetes (impaired fasting glucose) and its relationship with anthropometric indices among urban-dwelling university teachers. In addition, this study also reported on the prevalence of measurement based pre-hypertension and pre-diabetes

Conclusion And Recommendation
Despite the fact that this study adapted Europid cutoff to define central obesity, the findings of this study can alarm the epidemics of non-communicable diseases and obviate the need for Ethiopian cutoff to define central obesity. The expected rapid escalation of cardio-metabolic risks cannot be ruled out, particularly in Ethiopia, as it can destabilize economy which is still reeling from infectious diseases. However, it should be understood that anthropometric Europid cut-offs for detecting central obesity may not be appropriate for Ethiopians.
Previous studies outlined that cardio-metabolic risk factors disproportionately affect the less educated and poorer segment of the society which has been attributed to healthseeking behaviors (11,39,42). In contrast, a systematic review of studies from Sub-Saharan African countries (51) indicated that increased wealth and better education was associated with an increased risk of diabetes in both male and female participants. Given this contrasting suggestion from literature and the high number of pre-hypertension and pre-diabetes among university staff in our study, we recommend future researchers to conduct comparative studies on the prevalence and risk factors of cardio-metabolic risk factors between the less educated, economically disadvantaged rural and the highly educated urban populations in Ethiopia. We also suggest that appropriate screen methods are put in place to mitigate the public health crisis that may arise from undiagnosed hypertension and diabetes.

Consent for publication
Not applicable

Availability of data and material
Since this is funded work the raw data is the property of the University of Gondar and the data that are confidential cannot be made publicly available in order to protect participant's privacy. Data may be available to interested researchers upon formal request from the corresponding author (Balamurugan Janakiraman: bala77physio@gmail.com)

Competing interests
The authors declare that they have no competing interests and all authors also declare that they have no conflict of interest resulting from this work, not in financial or personal relationships.

Funding
This study was fully funded by University of Gondar scientific research grant number   Values are presented as mean (95% confidence interval) and frequency (percentage among each gender) adjusted for age. *Significant difference in mean BMI between age groups in men.  Values are presented as mean ± SD or N (%) adjusted for age as indicated. P < 0.05, (Chi square for categorical and analysis of variance for continuous variable). * P < 0.05 between non-obese and obese in the same group. BMI -body mass index, WC-waist circumference, WHtR-waist height ratio, WHR waist hip ratio, SBP systolic blood pressure, DBP diastolic blood pressure, HTN hypertension, FBG fasting blood glucose.