Performance of newly developed body mass index cut-off for diagnosing obesity among Ethiopian adults.

Background: Obesity is defined as unhealthy excess body fat, which increases the risk of premature mortality from noncommunicable diseases. Early screening and prevention of obesity is critical for averting associated morbidity, disability, and mortality. Ethiopia has been using the international (WHO's) BMI cut-off for diagnosing obesity even though its validity among Ethiopian population was questioned. To address this problem, a new body mass index cut-off was developed for Ethiopian adults using population-specific data. However, its performance in diagnosing obesity has not been validated. Therefore, this study determined the performance of the newly developed Ethiopian and World Health Organization (WHO) BMI cut-offs in detecting obesity among Ethiopian adults.

Methods: A cross-sectional study was carried out among 704 employees of Jimma University from February to April 2015. The study participants were selected using simple random sampling technique based on their payroll. Data on sociodemographic variables were collected using an interviewer-administered structured questionnaire. Anthropometric parameters including body weight and height were measured according to WHO recommendation. Body fat percentage (BF%) was measured using the air displacement plethysmography (ADP) after calibration of the machine. The diagnostic accuracy of the WHO BMI cut-off (≥ 30 kg/m²) for obesity in both sexes and Ethiopian BMI cut-off (> 22.2 kg/m² for males and >  24.5 kg/m² for females) were compared to obesity diagnosed using ADP measured body fat percentage (> 35% for females and >  25% for males). Sensitivity, specificity, predictive values, and kappa agreements were determined to validate the performance of the BMI cut-offs.

Results: In males, WHO BMI cut-off has a sensitivity of 5.3% and specificity of 99.4% (Kappa = 0.047) indicating a slight agreement. However, the Ethiopian cut-off showed a sensitivity of 87.5% and specificity of 87.7% (Kappa = 0.752) indicating a substantial agreement. Similarly, in females, the WHO BMI cut-off showed a sensitivity of 46.9%, while its specificity was 100% (Kappa = 0.219) showing a fair agreement. The Ethiopian BMI cut-off demonstrated a sensitivity 80.0% and a specificity 95.6% (Kappa = 0.701) showing a substantial agreement. The WHO BMI cut-off underestimated the prevalence of obesity by a maximum of 73.7% and by a minimum of 28.3% among males, while the values for underestimation ranged from 31.4-54.1% in females. The misclassification was minimal using the newly developed Ethiopian BMI cut-off. The prevalence of obesity was underestimated by a maximum of 9.2% and overestimated by a maximum of 6.2%. The WHO BMI cut-off failed to identify nearly half (46.6%) of Ethiopian adults who met the criteria for obesity using BF% in the overall sample.

Conclusions: The findings suggest that WHO BMI cut-off (≥ 30 kg/m²) is not appropriate for screening obesity among Ethiopian adults. The newly developed Ethiopian BMI cut-off showed a better performance with excellent sensitivity, specificity, predictive values, and agreement indicating the diagnostic significance of it use as a simple, cost-effective, and valid indicator in clinical and community setups.