Validity of Anthropometric Cut-Offs for Early Diagnosis of Dyslipidemia Among Ethiopian Adults

Introduction

Metabolic syndrome is a subset of non-communicable diseases (NCDs), which is interconnected by physiological, biochemical, clinical and metabolic factors that directly increases the risk of cardiovascular disease and type 2 diabetes mellitus. Insulin resistance, visceral adiposity, atherogenic dyslipidemia, endothelial dysfunction, genetic susceptibility, elevated blood pressure, hypercoagulable state and chronic stress are among several factors which constitute the syndrome.¹ Individuals with metabolic syndrome exhibit a characteristic pattern of abnormalities in serum lipid levels consisting of a low level of high density lipoprotein and an elevated level of triglyceride.²

Different studies show that the burdens of NCDs and their impact on the health of the population and health services are increasing in developing countries. Cardiovascular diseases, cancer, chronic respiratory diseases and diabetes are responsible for 82% of deaths. Currently, low income countries are witnessing epidemiological transition from infectious communicable diseases to chronic non-communicable diseases due to changes in lifestyles, rapid urbanization and diminishing levels of physical activity.^3,4 Consequently, non-communicable diseases are predicted to increase globally by 17% and by 27% in the African region in the next 10 years, and projected to overtake communicable diseases as the major cause of morbidity in sub-Saharan Africa by the year 2030.³

The 2017 WHO report progress monitor also showed that NCDs kill 41 million people each year, which is 71% of all global deaths. In Ethiopia, the proportion of death from NCDs has been reported at 42% of all causes of deaths, which accounts for 27% of premature deaths before the age 70 years.⁵ The increment of NCDs on top of the preexisting highly prevalent communicable diseases, maternal, prenatal and nutritional conditions constrains the already meager health resources and hinders economic development of the country.⁶

However, it is only relatively recently that NCDs have received attention in low-income countries. Consequently, unrecognized/undiagnosed NCDs are found to be major problems as the individuals are unaware of their status in this regard. It has been indicated that the international anthropometric cutoff for detecting obesity is not appropriate for Ethiopians.⁷ There is a need to reorient the national health system to ensure recognition of the NCDs burden and improve organization and delivery of NCDs services at primary health care (PHC) level.^8,9

Lack of local based cut-off points remains one of the challenges in management, prevention, and control of NCDs, and their modifiable risk factors. Studies show that the international cut-off points for different anthropometric measurements and body composition are not appropriate for the Ethiopian population; it leads to underestimation and misclassifies the risk of metabolic syndrome. Therefore, developing standardized local cut-off points for early detection of metabolic syndrome is crucial for prevention and management of NCDs. Thus, this study aimed to validate anthropometric cut-off points among Ethiopian adults⁷ through household level screening using waist circumference, waist-to-hip ratio and BMI measurement for estimation and classification of the risk of metabolic syndrome in Jimma Town, Oromia Regional State, Southwest Ethiopia.

Methods

Study Setting

A community-based cross-sectional study was conducted in Jimma Town from June to July 2019 among adults aged ≥18 years. Jimma is found in Southwest Ethiopia, located 352 km from the capital city of Addis Ababa. Sample size was calculated using sensitivity estimation formula¹⁰ taking highest prevalence of high density lipoprotein (23.4%) based on a study done in Addis Ababa among Ethiopian adults,¹¹ 5% margin of error, 95% confidence level and a 90% anticipated sensitivity.⁷ Finally, sample size of 977 was calculated by considering 10% non-response rate and 1.5 design effects. Study participants were selected using two-stage cluster sampling technique. Six kebeles (smallest administrative unit) out of 17 were randomly selected. Study participants were selected from each kebele proportionally based on the number of households in each selected kebele using systematic sampling technique.

Inclusion and Exclusion Criteria

All residents who lived in the town for six months prior to the data collection were included into the study. However, adults who had physical deformity (kyphosis and scoliosis), pregnancy, known chronic illness and serious illness were excluded.

Data Collection Procedure

Data were collected using WHO stepwise questionnaire and adapted to the local context. The survey tools included socio-demographic characteristics, anthropometric measurements and laboratory analyses of lipid profile. Height of the study participants was measured to the nearest 0.1 cm using a stadiometer (Seca GmbH & Co. KG, Germany) with the subjects positioned at the Frankfurt Plane and the four points (heel, calf, buttocks and shoulder) touching the vertical stand and their shoes taken off. Before starting the measurements, the stadiometer was checked using calibration rods. Weight was measured using digital weight scale to the nearest 0.1 kg with the subjects wearing light clothes and shoes taken off. The validity of the scale was checked using an object of a known weight of 1kg.

Waist circumference was measured at midpoint between the inferior margin of the last rib and the iliac crest just at wider area using a stretch tape. Just before taking the measurement, participants were requested to stand with their feet together, place their arms at the side of their body with the palms of their hands facing inwards, and breathe out gently. Hip circumference was measured at the level of the greater trochanter of the femur with the subjects wearing pants. All anthropometric measurements were done in triplicate and the average value was used for further analysis. Standardization of anthropometric measurers were done to reduce inter observer error.

The laboratory parameter was determined according to the standard operating procedures. Five 5 mL venous blood samples were collected to determine participant’s lipid profiles in mg/dl. Serum was carried out in ABX Pentra 400 Automated Chemistry Machine (Horiba ABX SAS, 34184) Montpellier¹² at Jimma medium Clinical chemistry Core laboratory to determine lipid profile including total cholesterol, high-density lipoprotein (HDL) and triglyceride (TG). Low density lipoprotein (LDL) level was calculated by using Freidwald formula.¹³

The optimal cut-off values of waist circumference (WC) was 85.3 cm for male and 78.0 cm for females and the values greater than this indicates risk factor for over-nutrition. The optimal cut-off values of waist-to-hip ratio (WHR) were calculated by dividing waist-to-hip in centimeters. The cut-off values were 0.88 cm for male and 0.82 cm for females; values greater than the cut-off value was considered at risk for over-nutrition. Similarly the optimal cut-off of values body mass index (BMI) was 24.5kg/m² for female and 22.2kg/m² for male; values greater than the cut off value was considered as at risk for over-nutrition. A lipid level was classified according to WHO classification and triglyceride was considered as normal when it was <150 mg/dl and HDL-C was considered as normal if it is >40 mg/dl male and >50mg/dl for female.¹⁴

Data Quality Management

Data collectors were trained for four days before the actual data collection on interviewing approach, anthropometric measurement and data recording. To ensure the quality of the data, a properly designed data collection instrument was used. Calibration of instruments was done to increase the reliability of the measurements. A questionnaire for the survey was first prepared in English language and translated into local languages (Afan-Oromo and Amharic), then back-translated into English to check for consistency. The questionnaire was pretested on 5% of sample size in Agaro Town and based on the pretest, additional adjustments were made based on identified gaps.

Data Processing and Analysis

Data were checked for completeness. The data were cleaned and checked for outliers and missing values. Epi Data version 3.1 and SPSS version 21 were used for data entry and analysis, respectively. The data were checked for normality using QQ-plot. Descriptive analysis was used to describe the study subjects and presented using tables and figures. Validity measures including sensitivity, specificity, positive predictive value, negative predictive value and kappa agreement were used to validate anthropometric based screening of lipid profile. The Kappa coefficient was used to measure the agreement between local anthropometric cut-off points and lipid profiles; and the kappa coefficient were classified as perfect (k=0.81–1.00), substantial (k=0.61–0.80), moderate (k=0.41–0.60), fair (k=0.21–0.40), slight (k=0.01–0.20) and no or poor agreement (k=0).^7,15

Results

A total of 915 individuals were included into this study with a 93.7% response rate. The mean age of study participants was 38.35±13.51 years. A large proportion of the respondents (71.5%) were married. The majority of participants were Oromo ethnic (66.0%) and Muslim followers (56.0%). Close to three-fourths (64.5%) of the study participants attended at least primary school (Table 1).

Table 1 The Socio-Demographic Characteristics of Adults in Jimma Town, Ethiopia

BMI-based screening of triglyceride in both sexes had slight Kappa agreement among female and fair agreement among males. Using BMI at ≥24 for screening of hyper triglyceride (TG ≥150mg/dl) among females showed slight (0.12) Kappa agreement (P=0.024) with 72.6% sensitivity and 65.0% specificity. Similarly, in case of male, BMI at ≥22.2kg/m² based screening of dyslipidemia (TG≥150mg/dl) had fair (0.275) Kappa agreement (P<0.001) with 64.1% sensitivity and 66.8% specificity. Body mass index-based screening of high-density lipoprotein (HDL-C<50mg/dl) among females at ≥24.5kg/m² had negative (−0.003) Kappa coefficient with 58.0% sensitivity and 41.7% specificity, which is not statistically significant (p=0.949). Similarly, BMI at ≥22.2kg/m² had a slight (0.032) Kappa coefficient (P=0.449) with 58.1% sensitivity and 45.45% specificity (Table 2).

Table 2 Comparison of BMI-Based Screening of Triglyceride and HDL with Laboratory Results Among Jimma Town Adults, Ethiopia

Waist circumference based screening of dyslipidemia (TG≥150mg/dl) at ≥78.0 among females had −0.005 Kappa coefficient with 72.6% sensitivity and 26.7% specificity, which was not statistically significant (p= 0.877). On the other hand, for males, waist circumference at ≥83.7cm based screening of dyslipidemia (TG≥150mg/dl) had slight Kappa agreement of 0.13 (p=0.005) with 38.0% sensitivity and 74.9%, specificity.

Waist circumference based screening of high-density lipoprotein cholesterol (<50mg/dl) at ≥78.0cm had slight Kappa coefficient (0.04) with 27.5% sensitivity and 77.3%, specificity which was not statistically significant (p=0.263). Similarly, for male waist circumference ≥83.7cm based screening of HDL-C had slight Kappa coefficient of 0.02, but not statistically significant (Table 3).

Table 3 Comparison of Waist Circumference-Based Screening of Triglyceride and HDL with Laboratory Results Among Adults, Jimma Town, Ethiopia

Waist circumference to hip ratio based screening of triglyceride was valid only for male at ≥0.88 cm with slight (0.105) Kappa coefficient (p=0.002) and 77.46% sensitivity and 36.83% specificity. However, in case females, WHR at ≥0.82cm was not statistically significant. WHR based screening of high-density lipoprotein cholesterol was not statistically significant for both female (p=0.263), and male (p=0.661) (Table 4).

Table 4 Comparison of WHR-Based Screening of Triglyceride and HDL with Laboratory Results Among Adults, Jimma Town, Ethiopia

Discussion

Developing locally appropriate optimal cut-off points using obesity based anthropometric measurement is important for early detection of metabolic syndrome to prevent farther complication and disabled from unrecognized NCD.⁷ According to the present study, BMI-based screening of triglyceride had fair and slight agreement of Kappa coefficient among males and females, respectively. This finding indicate that though the cut-off points for BMI is not the same for males (≥22.2kg/m²) and females (≥24 kg/m²) like the international cut-off points, it can be used to early detect dyslipidemia among Ethiopian adults. A similar study done in Brazil also revealed that anthropometric indices had slight differences in screening of triglyceride among males and females. This study also revealed that BMI-based screening of triglyceride was applicable when compared with other anthropometric measurements.¹⁶ Likewise, other hospital-based studies done in India, Canada and others indicated presence of correlation between BMI and lipid profiles.^17–19 However, in this study no significant association was observed between local BMI cut-off points and other lipid profiles like high and low density lipoprotein.

Waist circumference based screening of triglyceride among males showed slight agreement (k=0.13). However, similar studies from Canada and India found that waist circumference was a better predictor of dyslipidemia (TG) in both sexes.^18,20,21 Similar to waist circumference, the screening of triglyceride based on wait to hip ratio among males had slight Kappa agreement (k=0.105), which is statistically significant (p=0.002). Study done in India among adult population (≤ 60 years old) and aged group of population in Brazil (60 years old and above) revealed that anthropometric indices of waist to hip ratio was the best indicators for screening dyslipidemia (TG).^16,22 This discrepancy might be due to the difference in sample size, age difference or ethnic differences, which can be addressed by a large scale and multi-centered study involving different populations.

Limitations of the Study

Anthropometric measurements are prone to errors. However, this study was conducted with highest precautions in training the anthropometric data collectors, equipment calibration and standardization of procedure to minimize the possibility of error occurrence.

Conclusion

Anthropometric indices can be used as a simple screening for dyslipidemia at primary health care unit. BMI cut-off points at ≥24.5 kg/m² for females and ≥22.2kg/m² for males can be used for screening of dyslipidemia (TG≥150mg/dl). Similarly, waist circumference at ≥83.7cm and waist to hip ratio at ≥0.88 can be considered as screening of dyslipidemia (TG≥150mg/dl) among males. Generally, with a further investigation healthcare providers and researchers can use locally validated anthropometric indices as alternative to investigate lipid profiles to be used in the prevention and management of metabolic syndrome and other non-communicable diseases.