Development and validation of the body shape scale (BOSHAS) for assessing body shape perception in African populations

Subjects were recruited in the Niger–Congo region, Africa’s largest ethno-linguistic area spreading from Western to Southern Africa. Within this wide area, Bantoid (Western African ancestry group) and Bantu (Central African ancestry group) populations were selected, since they are known as linguistically distinct [38], phenotypically adapted to their contrasted ecological environment (lanky Western Bantoid in the Sahel vs. stocky Bantu in the equatorial forest) [39], and genetically different [40]. Bantoid and Bantu subjects were selected in Senegal and Cameroon, respectively [37].

The sample included 161 Senegalese and Cameroonian adult subjects (age range: 18–75 years), 31 males and 50 females from Dakar (Pikine area), and 51 males and 29 females from Yaoundé (Cité Verte area). The participants of the BOSHAS development protocol were recruited as part of a wider health survey to estimate the prevalence of obesity, diabetes and hypertension in both regions. Participants who were pregnant, < 18 years old or with a collapsed body posture from old age, were not recruited for this study because the aim was to collect photographs of adult bodies for the BOSHAS that did not belong to any specific age range within adulthood.

A dedicated room was rented in both cities during data collection. A Canon EOS 450D camera (Tokyo, Japan) was set up on a tripod height of 1 m, using an 18–55 mm lens. Using the maximum resolution of the camera (4272 × 2848 pixels), the front and the left side of participants were photographed. More technical details concerning the use of the camera and the position of participants are available in the development and validation of the BSS [27]. The mediolateral and the anteroposterior body axes were of particular interest to depict the variation in body shape. Indeed, this latter axis is rarely represented in existing body scales, even though its potential contribution to body image assessment studies [41, 42].

Subjects were measured using the body mass index (BMI), the body fat distribution and the somatotype standardized procedures to evaluate their morphology accurately. We used WHO BMI cut-offs to define underweight (< 18.5 kg/m²), normal-weight (18.5–24.9 kg/m²), overweight (≥25 kg/m²) and obesity (≥30 kg/m²) of subjects. Since this international BMI classification is not fully adapted to body composition variations across ethnic-groups, as in African populations [43], we complemented the BMI measurement with measures of body shape including body fat distribution and composition, as recommended by Adab et al. [44].

A digital beam scale (Tanita, Tokyo, Japan) was used to measured weight to the nearest 0.1 kg. A portable stadiometer (Siber-Hegner, Zurich, Switzerland) was used to measure standing height to the nearest mm. BMI (in kg/m²) was calculated by dividing weight (in kg) by squared height (in m). Hip, waist, upper arm (flexed and tensed) and calf circumferences were measured to the nearest mm in a standing position using a tape measure [45]. Waist-to-hip ratio (WHR) was calculated. A Harpenden skinfold caliper (Holtain Ltd., Crymych, UK) was used to measure triceps, supraspinal, subscapular and medial calf skinfold thicknesses to the nearest mm [46]. A Mitutoyo Dial Caliper (Mitutoyo America, Aurora, Illinois, USA) was used to measure the biepicondylar humerus and femur bone breadths.

Using the Somatotype 1.0 software package (MER Goulding Software Development, Geeveston, Australia), the somatotype of each participant was assessed from the calculation of its ternary component (musculature, slimness and fatness indexes) based on ten of the anthropometric measures described above [37], and stated in the BSS development protocol [27]. The somatocharts showing the somatotype distribution in our sample were also generated using this software. The assessment of the somatotype category was defined by combining the respective degrees of fatness (endomorphy), muscularity (mesomorphy) and slimness (ectomorphy) interpreted by a unit system from the Heath and Carter somatotyping metric [37]. The ternary components of each somatotype were plotted to represent subjects’ indices of musculature, slimness and fatness, respectively.

Ten subjects per sex were selected as models for the BOSHAS; this number provides an optimal trade-off between having a large enough choice with sufficient reliability [47]. We carefully ensured that we encompassed the anthropometric variability from the study population in terms of body shape (somatotype), adiposity (BMI) and body fat distribution (WHR). In our study population, we considered these three inclusion criteria simultaneously to select BOSHAS models. First, we randomly selected multiple clusters (i.e. small subsamples) within each of the main somatotype categories, as weakly endomorph, moderately mesomorph, strongly ectomorph. Second, we randomly chose a few individuals with the most contrasted BMI values from these clusters; and third, we retained individuals (the models) with the most opposite WHR for the BOSHAS within these last smaller clusters. Through this triangulation between somatotype, BMI and WHR, we retained different degrees of endomorph, mesomorph and ectomorph models for the BOSHAS, with both BMI and WHR variability, as explained below.

Based on the somatotyping metric, we selected five endomorph subjects for the female scale: one normal-weight (WHR: 0.86) strongly endomorph (“Endo++”: endomorphy is 0.5 unit higher than the two other components), two overweight (WHR: 0.86 and 0.95) and one obese (WHR: 0.80) moderately endomorph (“Endo+”: endomorphy is the first component but the second one is 0.5 unit higher than the third one) and one normal-weight (WHR: 0.68) weakly endomorph (“Endo-”: endomorphy is the first component but within a 0.5-unit range from the second component). We then selected one underweight (WHR: 0.75) relatively ectomorph subject (Ecto+) and one normal-weight (WHR: 0.82) central subject (all components are within a 1-unit range from each other). Finally, we selected three mesomorph subjects: one normal-weight (WHR: 0.78) strongly mesomorph (Meso++), one overweight (WHR: 0.74) and one obese (WHR: 0.84) relatively mesomorph (Meso+).

For the male scale, we selected two endomorph models: one overweight (WHR: 0.85) relatively endomorph (Endo+) and one overweight (WHR: 0.93) weakly endomorph (Endo-), one underweight (WHR: 0.81) strongly ectomorph (Ecto++), one normal-weight (WHR: 0.82) central subject, and six mesomorph subjects: one normal-weight (WHR: 0.87) strongly mesomorph (Meso++) and five relatively mesomorph (Meso+), one normal-weight (WHR: 0.88), one overweight (WHR: 0.97) and three obese (WHR: 0.84, 0.94, 0.98).

Adobe Photoshop CS software (Adobe Systems Inc., San Jose, CA) was used to edit the photographs, mask the faces, remove the shadows and introduce a white background.

A validation protocol was conducted in a random sample of Senegalese adults from Dakar (urban) and Kaolack (rural). In every 3rd household, male and female adults informed and aware of the study procedure and providing their agreement were invited to participate in the study.

Although the BOSHAS presents different human body shapes, these models are not arranged as decreasing or increasing body shape metrics. However, it is possible to validate this tool objectively according to the body shape preference of participants. For instance, participants pointed out different models for several body shape criteria (e.g. preference order, Current Body Shape (CBS)) at two different time points. This allowed us to determine the repeatability (test-retest) of their perceptions objectively. Hence the photographs were presented on large individual cards (width: 10 cm, length: 20 cm) so that the participants could accurately observe the body shape differences between models. The photos were also randomly numbered from 1 to 10 (the numbers were blinded on the back of the cards), shuffled at every question and presented to the participants blinded from any anthropometric information [47].

We assumed that the preference order of the models in the BOSHAS stated by participants constitutes a relevant dimension that could be used to evaluate the validity of this body image scale. Hence, we assessed the repeatability of the participants’ order to evaluate whether the BOSHAS allowed them to find their own hierarchical preference of the models. Accordingly, we asked the participants twice, with a test-retest interval of 2 weeks, to rank the BOSHAS models according to their preference order, a strategy of validation based on the classification of the models already applied for the BIS [28], the Body Size Guide (BSG) [48] and the Photographic Figure Rating Scale (PFRS) [49]. We recruited 106 Senegalese adults of both sexes (45 females and 61 males; age range: 18–77 years) who expressed their preference order on the BOSHAS for themselves as well as for their partner (for polygamous households, the wife was chosen by the husband). For own sex perception assessment, males were administered with the male scale and females with the female scale. For the partner perception assessment, the male and female scales were reversed. For both of the sex-specific subscales, we asked the following question in French or Wolof: “Peux-tu classer ces images de celle que tu préfères le plus à celle que tu préfères le moins? ” [Can you sort these models from your most to least preferred?]

As carried out for the PFRS and the BSS [27, 49], we administered the BOSHAS in the same sample using the protocol described earlier to evaluate the reliability of assessing current body shape (CBS and CBS’), desired body shape (DBS and DBS’) and ideal body shape (IBS and IBS’) for oneself and one’s partner. We evaluated the participant’s CBS by asking “Peux-tu montrer l’image qui te ressemble le plus?” [Can you point out the model that looks the most like you?]. We evaluated DBS by asking “Peux-tu montrer l’image à laquelle tu veux ressembler le plus?” [Can you point out the model that you most want to look like?]. We evaluated IBS by asking “Peux-tu montrer l’image idéale pour toi?” [Can you point out the ideal model for you?]. The difference between the DBS and IBS consists of the DBS means the reachable body shape expected by the participant and the IBS means the ideal body shape in absolute terms without the personal individual expectation [50]. We repeated the same protocol to assess perceptions of their partner’s body shape. We evaluated CBS’ by asking “Peux-tu montrer l’image qui ressemble le plus à ton partenaire?” [Can you point out the model that looks the most like your partner?]. We evaluated DBS’ by asking “Peux-tu montrer l’image à laquelle tu veux que ton partenaire ressemble le plus?” [Can you point out the model that you most want your partner to look like?]. We evaluated IBS’ by asking “Peux-tu montrer l’image de partenaire idéal pour toi?” [Can you point out the ideal partner model in your opinion?]. CBS/CBS’, DBS/DBS’ and IBS/IBS’ scores were defined as the number corresponding to the photo that the participant selected.

We recruited 597 adult Senegalese (age range: 18–101 years) living in Dakar agglomeration and Kaolack region (284 males and 313 females). The aim was to compare body shape perceptions assessed with the BOSHAS with those obtained by an interview-administered questionnaire, as examined in previous studies to test the convergent validity of the PFRS and the BSS [27, 51]. Usually, the methodological strategy consists of correlating dimensions assessed by a body scale (e.g. body dissatisfaction) with existing validated questionnaires measuring similar dimensions. However, our knowledge of the literature shows that no culturally relevant questionnaire on body image had been developed for African populations, and no body image questionnaire had been culturally adapted to African populations. The Body Shape Questionnaire might be an appropriate tool to be correlated with a body shape scale [50]. However, the cross-cultural adaptation of existing questionnaires is a complicated process [52], especially in the African context where large body shapes are still valued and food is perceived as a symbol of wealth [53], while body image questionnaires used in HIC populations tend to integrate slimness valorisation and eating disorders as norms. Traditional African aesthetics are not based on the rejection of the body and flesh, and individuals are still relatively protected from advertising and peer pressure to constantly improve their appearance, as observed in a recent qualitative socio-anthropological study in Senegal [54]. Therefore, we decided to develop an original culturally relevant questionnaire for this study (Table S2), based on the main findings of the previous socio-anthropological study, to specifically assess the convergent validity of the BOSHAS. We were able to create a simple questionnaire adapted to Senegalese body shape perceptions, which was recommended as an important criterion concerning the cross-cultural adaptation of existing questionnaires [55]. Accordingly, the following questions were administered by questionnaire:

“Sur un plan esthétique, vis-à-vis de votre poids, vous vous trouvez : 1. trop maigre, 2. un peu trop maigre, 3. normal, 4. un peu trop gros, ou 5. trop gros?” [“With regards to appearance, concerning your weight, do you feel: 1. too slim, 2. a bit too slim, 3. normal, 4. a bit too fat or 5. too fat?”].

For this item, we wanted to compare body weight self-satisfaction assessed by this question with that assessed by the BOSHAS.

“Quelle silhouette avez-vous selon vous : lutteur, charpenté, ventru, musclé, « sec » (pour les hommes), « taille-fine », « taille Coca-Cola », enrobée, fessue, forte (pour les femmes), moyen, mince, décharné, menu, gros, autre (pour les deux sexes)?” [“Which silhouette are you: wrestler, sturdy, potty, muscular, “muscular and thin” (for men), “slim-waist”, “Coca-Cola body”, coated, large-buttocks, sharp (for women), medium, slim, skinny, menu, big, other (for both)”].

For this item, we used common vernacular Wolof terms to qualify silhouettes, here translated into French and English. These terms were identified in a recent socio-anthropological study conducted in Senegal [54]. We compared the responses to this question to the BOSHAS-assessed CBS.

“Sur un plan esthétique, quelle place accordez-vous à votre apparence physique (silhouette, poids, traits, etc…) : 1. aucune, 2. très modérée, 3. modérée, 4. relativement importante, 5. importante?” [“With regards to aesthetics, what importance do you give to your appearance: 1. none, 2. slightly important, 3. moderate, 4. relatively important or 5. important?”].

“Vous voudriez être dans la norme de beauté moderne occidentale ou votre apparence physique importe peu?” [“Would you like to meet Western modern body norms or is your physical appearance not important for you?”].

We asked these two last questions because a socio-anthropological study [54] in Senegal shows that ‘Western’ acculturation through media promotes an increasing importance of physical appearance, not valued by traditional rural society, where one’s generosity and vitality are the most valued facets. We compared the responses from these items to the DBS, IBS and body self-satisfaction assessed by the BOSHAS.

We also compared the results from the BOSHAS to objective anthropometric measures [56]. We measured participants’ weight and height and we derived BMI classification to compare these anthropometric parameters to CBS (concurrent validity).

The characteristics of the three samples were described with means (± Standard Deviation, SD) and percentages. The somatotype analysis provided the somatotype profile of each subject. Frequency comparisons were then carried out to test the differences in somatotype status with regard to the country of origin, by using Fisher’s exact test.

For each sex-specific subscale, we calculated simple and weighted Cohen’s Kappas to compare the agreement between the first and second preference order of the participants. Test-retest reliability of CBS, DBS and IBS were evaluated using Spearman’s correlations and simple Cohen’s Kappas. We did not use weighted kappas, since the BOSHAS was not designed for use in a specific ascending/descending order.

We assessed concurrent validity of both sex-specific subscales by estimating the Spearman’s correlation between CBS (categorized by ascending BMI) and objective BMI as continuous. We also categorized BMI and CBS into four categories (underweight, normal weight, overweight and obesity) and evaluated their agreement using simple and weighted Cohen’s Kappas of both subscales.

We derived a body self-satisfaction variable from the BOSHAS: satisfied participants were those who pointed out the same silhouette for CBS and DBS, while the others were considered not satisfied [57]. We compared the BOSHAS-derived self-satisfaction to the body weight self-satisfaction as declared by questionnaire (variable 1: normal vs. other modalities) through the sensitivity and specificity estimations as in previous studies [58‐60].

We derived a “normal” body self-assessment variable from the BOSHAS corresponding to all silhouettes as perceived CBS valued in Senegalese society: ‘Coca-Cola’ body (model 5: a thin women with large buttocks and big breasts) and medium body (models 2, 3 and 5) in women, and wrestler and muscular (model 4, 7 and 10) and medium (model 1, 4, 7 and 8) bodies in men. A medium body on both scales corresponds to models between normal-weight bodies that are a bit fleshy, and overweight bodies without a big belly, for both sexes. These morphologies (1, 4, 7, 8 for men; 2, 3, 5 for women) represent vitality for the Senegalese beyond recent differences noted between perceptions based on age and associated with ‘Western’ acculturation [54]. On BOSHAS, we coded all models that corresponded to these silhouettes as normal CBS. We tested whether normal CBS predicted the self-assessed normal body on the questionnaire (variable 2: “normal” silhouettes as wrestler, Coca-Cola body, medium vs. others) by estimating sensitivity and specificity. We also tested whether BOSHAS-derived self-satisfaction predicted the normal body (variable 2).

Finally, we derived modern DBS/IBS variables from BOSHAS. Modern DBS/IBS corresponded to models 4 and 10 on the male scale and 3 and 5 on the female scale. Indeed, young people (possibly influenced by ‘Western’ media) value muscular bodies in men, and “sexy” bodies in women, defined as having large buttocks and breasts, but no fat. The value of these exigent aesthetic criteria could lead to body dissatisfaction [8]. Therefore, we tested whether BOSHAS-derived self-dissatisfaction predicted the importance of appearance (variable 3: relative/important vs. other modalities). Similarly, we tested whether the BOSHAS-derived modern DBS/IBS and self-dissatisfaction predicted modern aesthetic criteria (variable 4: modern vs. no modern).

Statistical analyses were performed using STATA 12 (StataCorp, College Station, TX, USA).

Selected Cameroonian males were more likely to be mesomorph and Senegalese males more likely to be ectomorph; frequency of endomorphy was similarly low in both groups (p < 0.01; Table S1). Similar, but less pronounced trends in ectomorphy and mesomorphy were found between Cameroonian and Senegalese females; however, in contrast to males, the frequency of endomorphy was similarly high in both groups (p < 0.05).

Figures 1 and 2 show the sex-specific subscales of real African models photographed from both face and profile, containing the 10 males/10 females selected as models for the BOSHAS, together with their corresponding anthropometric data. The models selected are not arranged in any specific order based on anthropometric characteristics but cover the somatotype variability of the initial sample. The wide variability of somatotype components in the BOSHAS for each sex is presented in the two somatocharts as Figs. 3 and 4. Secondly, the BOSHAS also integrated BMI and WHR variability within somatotype variability.