Neck circumference as marker of overweight and obesity among secondary adolescents aged 12-19 years in southwestern Uganda: a cross-sectional survey

Overweight and obesity have become more common among adolescents. Various indicators have been used to assess this burden across populations. Recent ndings elsewhere have found neck circumference to be useful in dening overweight and obesity. However, the use of neck circumference as marker of overweight and obesity needs to be further explored among adolescents in Uganda. Objective To assess the usefulness of neck circumference as a marker of overweight and obesity. A cross sectional study was conducted among secondary adolescents aged 12–19 years from May to August in 2018, Height was measured using a wall mount height board and weight using a standard weighing scale. Neck circumference was measured using an inelastic measuring tape. We dened overweight as BMI = 25.0-29.9kg/m 2 while obesity as BMI ≥ 30.0kg/m 2 among the study participants. Pearson correlation coecient was used to determine the correlation between neck circumference, height, weight and body mass index. Receiver operating characteristic (ROC) curve analysis was used to determine the suitable cutoff of neck circumference for overweight and obesity.


Introduction
Overweight and obesity, according to the world health organization (WHO), are de ned as excessive accumulation of fat that may impair health. Obesity and overweight are a global problem, and in 2016 alone, according to WHO, over 340 million children and adolescents aged 5-19 years were overweight or obese. The burden of obesity and overweight is highly linked to the incidence of non-communicable diseases ranging from diabetes mellitus, cardiovascular disease (CVD) [1] among others. Several anthropometric indices have been proposed to ease the identi cation of overweight or obesity. Among these, body mass index (BMI) is the most commonly used anthropometric index [2]. Several others such as waist circumference (WC), waist hip ratio (WHR), neck circumference (NC) and waist height ratio (WHtR) have also been proposed [3][4][5]. Neck circumference, has of recent attracted attention in health research and clinical settings. It has been reported to have potential of accurately de ne overweight and obesity across several populations [6][7][8][9][10][11][12], yet it is easy to measure. A cross-sectional study conducted in Indonesia that involved up to 1554 participants found high prevalence (79.7%) of obesity. In the same study, NC was found to be signi cantly correlated with BMI and WC, and was found to be useful for screening for overweight and obesity [7]. The study proposed cutoffs of 33.5 cm among the females and 37 cm among the males with high value of speci city and sensitivity. Among Bangladesh adults, NC was found to have the ability to distinguish overweight and obese adults from their counterparts [10]. Findings from a study that included 864 children and adolescents from Iran, showed that NC was signi cantly correlated with BMI, WC and Mid upper arm circumference (MUAC) [13]. The proposed NC cutoffs for overweight or obesity were 27.5-38.3 cm in boys versus 26.7-33.4 cm in girls. All the ndings give highly credible data on the usefulness of NC in screening for overweight or obesity. However, other studies are needed to localize nding to Uganda's population with emphasis on adolescents, with overweight and obesity currently affecting 12% of the Ugandan adolescents and young adults of 12-24 years [14]. We aimed to assess the usefulness of neck circumference as a marker of overweight or obesity.

Methods
This was a cross-sectional study conducted among secondary school adolescents in Mbarara Municipality, southwestern Uganda from May to August 2018.

Data collection
Age and sex of the participants was self-reported in complete years. Neck circumference (NC) was assessed as a surrogate measure for upper body adipose tissue distribution. It was measured at the level of the laryngeal prominence using an inelastic exible measuring tape, with the subjects in the standing position, and the head held erect and eyes facing forward to the nearest 0.1cm [9,15].
Height was measured in centimeters with the participant shoeless and rear body parts touching the board while the head is facing forward, using a wall mount height board without shoes [15][16][17].
Weight was measured to the nearest 0.5kg using a standard weighing scale (Seca 762, GmbH & Co. KG, Hamburg, Germany). The participants were encouraged to put on light clothes and without items in the pocket [15,18]. The participants were not allowed to put on shoes.
Body mass index (BMI) was calculated as the ratio of the weight of the participant in kilograms to height in square meters [19]. Overweight was de ned as BMI =25.0-29.9kg/m 2 while obesity as BMI ≥30.0kg/m 2 .

Statistical analysis
Data were collected in excel form powered by open data kit, and were analyzed using Stata 13.0 (College Station, Texas, USA). All data were analyzed and reported according sex category. The means and standard deviations (SD) were used to describe continuous variables while for the categorical data, frequencies and percentages were estimated. The associations between NC and other anthropometric variables of the studied participants were assessed using Pearson's correlation analysis. Receiver operating characteristic (ROC) curve analysis was used to determine optimal sex-speci c cutoffs of NC in relation to BMI. The cutoff was considered as the value of NC that showed the maximum Youden's J statistic [20]. Area under the curve (AUC) and its 95% con dence interval were reported. Statistical signi cance was set at p< 0.05.

Characteristics of study population
The baseline characteristics of study participants have been described in detail elsewhere [15]. Brie y, data of 616 adolescents, of which 212 were males was analyzed. The mean age of the participants was Distribution of BMI categories among study participants Among our study participants, the prevalence of obesity was higher among the females at 4.5% (n=18) in comparison to 1.4% (n=3) among the males. The overall prevalence of obesity was 3.4%. as shown in table 1.

Correlation of neck circumference with other obesity indicators
Overall, there was a signi cant (p<0.001) positive Pearson correlation coe cient between NC and height (r=0.56, P<0.001); weight (r=0.55, P<0.001); and BMI (r =0.19, P<0.001), as shown in Table 2. After stratifying for sex, all correlations were positive and signi cant among the females. Among the males, the correlation between NC and BMI was positive but not signi cant.

Receiver operating characteristic (ROC) curve analysis
Among the males, a cut off of NC ≥ 29.8 cm was optimal for de ning overweight with an AUC of 0.60[0.54-0.66] as shown in Figure 1. In Figure 2, the cutoff of obesity among the males was determined as NC ≥ 32.0cm with a recorded AUC of 0.59 [0.11-1.00]; however, this was low and had a wide con dence interval overlapping with the null as shown in Table 3. Among the females, a cutoff of NC ≥ 29.8cm was appropriate to indicate overweight with a recorded AUC of 0.5993 as shown in gure 3, while NC ≥ 31.0 cm indicated obesity with a very high AUC of 0.83[0.73-0.94], with of positive predictive values and negative predictive values of 0.778 and 0.741 respectively.

Discussion
The ndings from our study, revealed the usefulness of measuring NC to screen for overweight and obesity among adolescents in Uganda. A positive correlation between NC and BMI was generally reported. We also determined the reliable cutoff values for overweight and obesity in both sexes, with NC being more useful and accurate among the girls in de ning overweight and obesity. This could be attributed to the fact that AUC depends greatly on the prevalence of the disease, given that obesity was more among the females (4.5%) than the males (1.4%) in our study population. Additionally, our study ndings have given a green light to large nationwide surveys to assess the usefulness of NC in de ning overweight or obesity. Our ndings are in agreement with several studies that reported positive signi cant correlations between NC, and other indicators of overweight or obesity. Among 1102 children and adolescents, a positive correlation was reported between NC, BMI and WC among others [21]. The same study proposed NC cutoffs among boys ranging from 28 to 39.0 cm, while among the girls, the cutoffs ranged from 27.0 to 34.6cm. Furthermore, the same and related ndings were reported among 412 adolescents aged 6-18 years that were recruited from the Endocrinology department of Erciyes University in Turkey [22]. The same ndings were reported in Egypt, by a study that involved overweight and obese adults [23]. The other studies reporting the same related ndings include [7,24]. Our study, was the rst of a kind to report such results from adolescents in Uganda. Based on our ndings, we recommend that an extensive and highly inclusive study be conducted to determine the overall cutoffs and further exploitation of the relationship of NC with metabolic syndrome and its components such as hyperlipidemia, diabetes and hypertension. However, despite of the results, there are several limitations to our ndings. The study was considered only in southwestern Uganda and only among secondary school adolescents in Mbarara municipality, which is more urbanized municipality. This makes results generalization less likely. Additionally, our analysis was based on the AUC from ROC analysis and this is generally affected by the prevalence of the disease among the study population. Thus, the results here may not apply to areas with very high prevalence of overweight or obesity.

Conclusion
Neck circumference is an easy and noninvasive marker of obesity among the females in our study population compared to males. It can be used for screening for obesity among only the female adolescents. Availability of data and materials Data and materials ware available on request from the corresponding author.

Competing interests
The authors declare no competing interests.

Funding
The study received no external funding.
Authors' contributions GK, DCA and RM: Conceptualization of work & its realization, wrote the manuscript, checked the references, compiled the literature sources, data collection, statistical analysis, and interpretation of data, and wrote the manuscript. FM, HM and RN: mentored the conceptualization of work & its realization, compiling literature sources and statistical analysis, helped in data interpretation, guided manuscript writing, checked the references. AM, KAM and AN: helped in the conceptualization of the work, helped in statistical and data analysis, support to data collection. All authors read and approved the manuscript before submission.    Figure 1 Receiver operating curve analysis of neck circumference with male overweight participants (BMI =25.0-29.9kg/m2). NC cut off ≥29.8cm Receiver operating curve analysis of neck circumference with male obese participants (BMI ≥30.0kg/m2). NC cut off ≥32.0