Sex Differences in Diet and Physical Activity Behaviors Among Racial/Ethnic Minority Adolescents with High Metabolic Risk

Abstract

Certain dietary and physical activity (PA) behaviors may differentially predispose male and female adolescents to obesity and diabetes; however, sex differences in dietary and PA behaviors and in factors that impact these behaviors (e.g., self-efficacy, social support) in this population remain unknown. Using data from a community-based adolescent diabetes prevention intervention conducted in East Harlem in New York City, we examined sex differences in baseline characteristics including clinical measurements, lifestyle behaviors, and behavioral determinants. Among 147 overweight/obese adolescents aged 13–19 years, 61.9% were girls, 69.7% were of Hispanic ethnicity, 24.8% were non-Hispanic Black, and 60.5% were diagnosed with prediabetes. Boys had higher metabolic risk scores than girls (3.8 vs. 3.3, p = 0.002) despite girls reporting more perceived barriers to healthy eating and PA. Boys reported doing more moderate to vigorous PA but also had more sedentary behaviors than girls. Boys reported higher self-efficacy and more peer support for PA. Girls reported more depressive symptoms and were more likely to compare their body images to those in magazines/social media. Overall, among a sample of urban adolescents with high metabolic risk, we found significant sex differences in many dietary and PA behaviors and related factors, which could be used to inform tailored strategies for weight management to reduce cardiometabolic risk among youth from similar high-risk populations.

Introduction

Childhood obesity is a significant public health problem in the USA, putting children and adolescents at risk for short-and long-term comorbidities, such as diabetes, hypertension, obstructive sleep apnea, dyslipidemia, nonalcoholic fatty liver disease, orthopedic conditions, and depression [1–4]. According to a US National Center for Health Statistics report, obesity prevalence for youth ages 6–11 years and 12–19 years steadily increased from approximately one in every 25 children from 1963 to 1965 to one in every five children from 2017 to 2018 [5]. Disparities in obesity prevalence by race/ethnicity and age are well documented, with higher prevalence among Hispanic/Latinx and African American/Black children than among non-Hispanic white children and higher prevalence among adolescents than among younger children [1, 5–8]. However, few studies have examined sex differences in obesity rates and factors influencing metabolic risk.

The few studies that have examined sex differences in obesity and related behaviors have reported mixed findings. For example, previous studies revealed no sex differences in obesity rates among children and adolescents [9, 10]. However, one study found that boys had higher physical activity (PA) self-efficacy and higher PA levels than girls, with a stronger relationship between self-efficacy and PA among girls [11]. Prior studies have also found differences in weight control behaviors with girls reporting more dieting, use of vomiting/laxatives, and binge eating than boys [12]. In addition, body dissatisfaction has been associated with obesity, with the adverse consequences of obesity including low self-esteem, depression, and body dissatisfaction, playing a pivotal role in the development of eating disorders, particularly in teenage girls [13–16].

Further examination of sex differences in clinical outcomes and lifestyle behavior factors, especially among populations disproportionately impacted by obesity and diabetes including racially and ethnically minoritized adolescents, may be important as many of these factors are modifiable, and study findings may inform targeted interventions. In this study, we examined these outcomes of interest among a group of diverse adolescents who were recruited as part of an adolescent diabetes prevention study in New York City.

Methods

We utilized baseline data from a randomized controlled trial of a diabetes prevention intervention targeting diverse adolescents with prediabetes from East Harlem in New York City. East Harlem residents are predominantly Hispanic/Latinx and/or Black and have among the highest obesity prevalence and diabetes mortality rates in New York City [17]. The study, named TEEN HEED (Help Educate to Eliminate Diabetes), utilized community-based participatory research (CBPR) strategies to develop, implement, and evaluate outcomes of a peerled diabetes prevention program. Detailed study design and methods have been described previously [18]. For this study, we analyzed baseline sex differences in clinical outcomes, diet and physical activity behaviors, and related factors.

We recruited adolescents from collaborating community youth organizations, health centers, and schools between November 2016 and September 2018. Inclusion criteria included English-speaking adolescents (with English- or Spanish-speaking caregivers) aged 13–19 years who were affiliated with the East Harlem community (lived in, went to school in, received health care in, or attended a community-based program in East Harlem) and who were overweight/obese (body mass index (BMI) ≥ 85th percentile based on published CDC growth curves) [19]. A total of 149 participants attended the baseline study visit, and 147 had complete data and were included in the analysis.

We obtained informed consent including written parent/guardian consent and participant assent for those < 18 years and participant consent for those ≥ 18 years. The study protocol was approved by the institutional review board at the Icahn School of Medicine at Mount Sinai.

Measures

Trained research staff completed clinical measurements in person during the baseline study visit. We collected data on demographics (age, race/ethnicity, parental birthplace (USA vs. outside of the USA), parental educational attainment, and language used at home), medical history (high blood sugar, high blood pressure, asthma, high cholesterol, sleep apnea, etc.), history of diabetes in a family member or close contact, self-reported health status, access to care, and physician advice related to diet and physical activity behaviors. We measured height with a stadiometer (Quick Medical #14,268) and weight and body composition using a Tanita bioimpedance analyzer scale (Competitive Edge #SC-3315). BMI percentile and Z-score were calculated based on measured height and weight using the SAS program by the Centers for Disease Control [19]. We measured systolic and diastolic blood pressure using an ambulatory blood pressure monitor (Quick Medical BPM-200), recorded the average of the final two of three readings, and used the National Heart, Lung, and Blood Institute (NHLBI) blood pressure guidelines to record blood pressure percentiles [20, 21]. The average waist circumference was calculated based on two measurements obtained with a Gulick tape measure (Model 67,019) using a standard protocol. We obtained fasting glucose and glucose 2 h after a 75-g oral glucose load by fingerstick and analyzed with a glucometer (Henry Schien #1,068,963). Fasting dried blood spot testing was used to measure the following additional clinical measurements: (1) lipid panel, including total cholesterol, LDL, VLDL, HDL, and triglyceride levels; (2) insulin level; (3) hemoglobin A1c; and (4) highsensitivity C-reactive protein level. All blood samples were collected and frozen at −80 °C before analysis at a Clinical Laboratory Improvement Amendments (CLIA)–certified laboratory using validated dried blood spot testing procedures (https://www.zrtlab.com/sample-types/blood-spot/) (ZRT labs; Beaver, OR). Finally, we calculated a previously validated summary metabolic risk score (MetS) for each participant based on their sex, race/ethnicity, BMI Z-score, HDL, triglycerides, systolic blood pressure, and fasting glucose [22].

Trained interviewers verbally administered the survey to ensure accuracy of the results and enable those with limited written literacy to participate. We developed our survey collaboratively with our Community Action Board to assess domains in our diabetes prevention conceptual model including knowledge, attitudes, beliefs, behaviors, cognitive mediators, and demographic factors [23–25]. Survey domains and sources of questions are presented in Supplemental Table 1. The survey primarily focuses on factors related to diet and PA, such as portion control, perceived benefits and barriers to a healthy lifestyle, self-efficacy, and social influences. Research staff input survey data into REDCap, a secure web application for building and managing online surveys and databases (http://project-redcap.org/).

Statistical Analysis

We used frequencies and proportions to summarize descriptive statistics for categorical variables and median and inter quartile range (IQR) for continuous variables. We compared sex differences in sample characteristics using Chi-square for categorical variables (Fisher’s exact tests when the sample size was small) and Wilcoxon rank-sum tests for continuous variables. The p-value significance level was set at α < 0.05. The analysis was conducted in SAS version 9.4 (SAS Institute Inc., 2013.).

Results

Of the 147 overweight/obese adolescents, 61.9% were girls, 69.7% were of Hispanic/Latino(a) descent, 24.8% were non-Hispanic Black, 34.5% reported mainly speaking Spanish in their home, 55.9% reported that their parents were foreign-born, and 81.4% reported that their parent(s) graduated from high school (Table 1). With regard to medical history, 35.2% of boys reported being told by a physician that they had high blood pressure compared to 11.0% of girls (p = 0.0004), after adjusting for multiple testing. There were no significant sex differences in age, race/ethnicity, parental birthplace, parental educational level, or primary language spoken at home, following adjustments for multiple testing.

Table 1.

Sex differences in study population characteristics: demographics and medical history show in n (%)

	Total (n = 147, 100%)	Boy (n = 56, 38.1%)a	Girl (n = 91, 61.9%)	p-valueb
Demographics
Age				0.221
13–15 years	71 (49.0)	30 (55.6)	41 (45.1)
16–19 years	74 (51.0)	24 (44.4)	50 (54.9)
Race/ethnicity				0.0476
Black or African American	36 (24.8)	10 (18.5)	26 (28.6)
Hispanic or Latino/a	101 (69.7)	38 (70.4)	63 (69.2)
Others (NH-White, Asian, mixed, etc.)	8 (5.5)	6 (11.1)	2 (2.2)
Parental country of birth				0.687
USA (including Puerto Rico)	64 (44.1)	25 (46.3)	39 (42.9)
Other country	81 (55.9)	29 (53.7)	52 (57.1)
Parent education attainment				0.350
Did not finish high school	27 (18.6)	8 (14.8)	19 (20.9)
High school or higher	118 (81.4)	46 (85.2)	72 (79.1)
Main language spoken at home				0.784
English	86 (59.3)	33 (61.1)	53 (58.2)
Spanish	50 (34.5)	17 (31.5)	33 (36.3)
Other	9 (6.2)	4 (7.4)	5 (5.5)
Medical history
History of high blood sugar	85 (57.8)	25 (46.3)	60 (65.9)	0.020
History of high blood pressure c	29 (19.7)	19 (35.2)	10 (11.0)	0.0004
History of asthma	44 (29.9)	20 (37.0)	24 (26.4)	0.177
History of high cholesterol	24 (16.3)	12 (23.1)	12 (13.2)	0.128
History of sleep apnea	8 (5.4)	3 (5.4)	5 (5.5)	1.000
Diagnosed with condition that makes it hard to do age-appropriate things	20 (13.6)	12 (22.2)	8 (8.8)	0.023
Family history
Family history of diabetes	104 (70.7)	40 (74.1)	64 (71.1)	0.701
Close contact with diabetes	42 (28.6)	12 (22.6)	30 (33.0)	0.189
Perceived health status (self-reported general health)
In general, would you say your health is excellent/good?	66 (44.9)	29 (53.7)	37 (40.7)	0.269
Medical home (usual access to care)
There is a particular doctor, medical person, or clinic for usual care	120 (81.6)	42 (77.8)	78 (85.7)	0.221
Easy to get medical care when needed	132 (89.8)	48 (90.6)	84 (92.3)	0.715
Physician advice
Advised to lose weight	118 (80.3)	46 (85.2)	72 (79.1)	0.365
Advised to eat healthier	133 (90.5)	52 (96.3)	81 (89.0)	0.568
Advised to get more exercise	126 (85.7)	46 (85.2)	80 (87.9)	0.638

^aThere were two missing observations in the male group

^bChi-square is used for categorical variables (Fisher’s exact test was used when the sample size was small [n < 5]). Wilcoxon rank-sum test was used for continuous variables

^cResult was significant after adjustment for multiple testing using Bonferroni test

Table 2 presents sex differences in clinical measures. More than 60% of all participants were diagnosed with prediabetes (fasting plasma glucose 100–125 mg/dl and/or glucose 2 h after an oral glucose load 140–199 mg/dl), with no difference based on sex (p = 0.974). Boys had a higher metabolic risk score than girls (3.8 vs. 3.3; p = 0.002), whereas girls had higher glucose levels 2 h after administration of an oral glucose load (127 mg/dl vs. 106 mg/dl in boys, p = 0.003). Girls had a higher body fat percentage than boys (41% vs. 30% in boys, p < 0.0001). There were no differences between boys and girls in fasting glucose level, BMI Z-score/percentile, waist circumference, systolic and diastolic blood pressure percentiles, insulin level, or lipid levels.

Table 2.

Sex differences in clinical measures in median (lower quartile, upper quartile)

Variable	Normal range	Boy (N = 56)	Girl (N = 91)	p-value*
Clinical measurements
Diagnosed with prediabetes	-	34 (60.7%)	55 (60.4%)	0.974
Fasting glucose level (mg/dl) [50]	70–100	102 (92, 109)	101 (92, 109)	0.915
Two hours after oral glucose (mg/dl) [50]	< 140	106 (90, 124.5)	127 (100.5, 141.5)	0.003
BMI Z-score [51]	− 1.64–1.04	2.1 (1.5, 2.5)	2.0 (1.6, 2.2)	0.391
BMI percentile [51]	5–85	98 (94, 99)	97 (95, 99)	0.381
Body fat percent [52]	10.1–20.3 in boys, 15.5–30.1 in girls	30 (22, 39)	41 (38, 46)	< 0.001
Waist circumference (cm) [53]	57–99	97 (87, 109)	95 (87, 106)	0.916
Systolic blood pressure percentile (%) [54]	Age and height related, < 95 in general	27 (12, 63)	40 (16, 60)	0.539
Diastolic blood pressure percentile (%) [54]	Age and height related, < 95 in general	52 (36, 71)	63 (39, 73)	0.198
Total cholesterol (mg/dl) [55]	< 200	165 (140, 194)	162.5 (144, 178)	0.745
LDL cholesterol (mg/dl) [55]	< 130	90 (73, 103)	88 (75.5, 103)	0.730
VLDL cholesterol (mg/dl) [56]	2–30	22 (14, 35)	22 (17, 30)	0.985
HDL cholesterol (mg/dl) [55]	> 45	46 (38, 60)	49 (40, 64)	0.352
Triglycerides (mg/dl) [55]	< 130	112 (71, 175)	110 (84, 150)	0.994
Insulin (μIU/ml) [57]	2–20, girls higher than boys	14 (10, 24)	18 (12, 26)	0.043
HbA1c [58]	< 5.7%	5.2 (4.6, 5.9)	5.1 (4.7, 5.8)	0.914
hsCRP level (mg/L) [59]	< 3	0.7 (0.4, 1.9)	1.2 (0.6, 2.8)	0.078
Metabolic score (MetS)a	-	3.8 (3.2, 4.2)	3.3 (2.7, 3.7)	0.002

^*Chi-square is used for categorical variables. Fisher’s exact test was used when the sample size was small (< 5), and Wilcoxon rank-sum test was used for continuous variables

LDL low density lipoprotein, VLDL very low density lipoprotein, HDL high density lipoprotein, hsCRP high-sensitivity C-reactive protein

^aMetabolic score is calculated based on Gurka et al. with the following formula: Non-Hispanic White male:

MetS = 0.2804 × body mass index (BMI) Z-score + 0.0257 × high density lipoprotein cholesterol (HDL) + 0.0189 × systolic blood pressure (SBP) + 0.6240 × ln(triglyceride) + 0.0140 × glucose(Glu) − 4.9310 Non-Hispanic Black male:

MetS = 0.2401 × BMI Z-score + 0.0284 × HDL + 0.0134 × SBP + 0.6773 × ln(Tri) + 0.0179 × Glu − 4.7544 Hispanic males:

MetS = 0.2930 × BMI Z-score + 0.0315 × HDL + 0.0109 × SBP + 0.6137 × ln(Tri) + 0.0095 × Glu − 3.2971 Non-Hispanic White female:

MetS = 0.4849 × BMI Z-score + 0.0176 × HDL + 0.0257 × SBP + 0.3172 × ln(Tri) + 0.0083 × Glu − 4.3757 Non-Hispanic Black female:

MetS = 0.5136 × BMI Z-score + 0.0190 × HDL + 0.0131 × SBP + 0.4442 × ln(Tri) + 0.0108 × Glu − 3.7145 Hispanic female:

MetS = 0.3520 × BMI Z-score + 0.0263 × HDL + 0.0152 × SBP + 0.6910 × ln(Tri) + 0.0133 × Glu − 4.7637

Table 3 presents diet, PA, and weight control behaviors, attitudes and factors impacting these behaviors, and significant differences in these behaviors and behavioral determinants.

Table 3.

Sex differences in diet and physical activity behaviors and behavioral determinants shown as N (%) or median (lower quartile, upper quartile)

	Boy (N = 56)	Girl (N = 91)	p-value*
Attitudes about diabetes and health (Agree Or Strongly Agree)
I am very concerned about getting health problems related to diabetes	45 (80.4)	84 (92.3)	0.032
I am worried about getting diabetes	39 (69.6)	82 (92.1)	0.0004
It’s not difficult to change diet and exercise to prevent diabetes	33 (58.9)	30 (33.0)	0.002
General dietary behaviors (in the past month eating 3 or more times per week)
I have eaten extra meals, snacks, or “seconds”	21 (37.5)	17 (18.7)	0.036
Portion control
Portion control scale (median (lower quartile, upper quartile)) (possible range [8–32], higher score indicates better performance on portion control)	21 (20, 23)	22 (19, 25)	0.291
I order large or extra-large when you eat fast food (such as pizza, Chinese, hamburgers, fried chicken) (Usually or Always)	7 (12.5)	2 (2.0)	0.027
Self-efficacy diet
Healthy lifestyle self-efficacy scale: diet (median (lower quartile, upper quartile)) (possible range [8–40], higher score indicates higher self-efficacy for healthy eating)	26 (22, 30)	25 (20, 29)	0.573
I can eat healthy foods when hungry after school (4 or 5 on 5-point Likert scale)	37 (66.1)	41 (45.1)	0.013
I can eat healthy foods when bored (4 or 5 on 5-point Likert scale)	27 (48.2)	25 (27.5)	0.011
I can change or maintain eating patterns to limit eating at fast food restaurants to once per week or less (4 or 5 on 5-point Likert scale)	49 (87.5)	66 (72.5)	0.033
Healthy eating barriers
Perceived barriers scale: diet (median (lower quartile, upper quartile)) (possible range [15, 60], higher score indicates more perceived barriers to healthy eating)	35 (31, 38)	37 (33, 41)	0.017
I eat unhealthily when in a bad mood (Agree/Strongly Agree)	17 (30.4)	49 (53.9)	0.005
It’s hard to eat healthy when hungry (Agree/Strongly Agree)	20 (35.7)	48 (52.8)	0.044
Unhealthy foods are too tempting (Agree/Strongly Agree)	23 (41.1)	62 (68.1)	0.001
Healthy foods are too expensive (Agree/Strongly Agree)	18 (32.1)	46 (50.6)	0.029
I am too rushed in the morning to eat a healthy breakfast (Agree/Strongly Agree)	23 (41.1)	64 (70.3)	0.0005
Eating healthy meals takes too much time (Agree/Strongly Agree)	5 (8.9)	20 (22.0)	0.041
Perceived benefits of healthy eating
Perceived benefits scale: diet (median (lower quartile, upper quartile)) (possible range [5–20], higher score indicates more perceived benefits of healthy eating)	15 (13, 17)	15 (14, 18)	0.142
Home food environment
Healthy food at home scale (median (lower quartile, upper quartile)) (possible range [8–32], higher score indicates greater access to healthier food at home)	17 (15, 20)	17 (14, 20)	0.836
Unhealthy food at home scale (median (lower quartile, upper quartile)) (possible range [5–20], higher score indicates greater access to more unhealthy food at home)	9 (7, 11)	10 (7, 13)	0.112
Physical activity
Moderate to vigorous physical activity scale (total hours reported per week (median (lower quartile, upper quartile))	6.6 (2.6, 8.6)	4.6 (1.6, 6.6)	0.004
Two or more hours of strenuous exercise (heart beats rapidly) in the past week. Examples: biking fast, dancing fast, running, jogging, swimming laps, rollerblading, soccer, basketball	33 (60.0)	31 (34.1)	0.002
I usually bike, skateboard, scooter, or rollerblade for more than 20 blocks in a day	13 (23.6)	6 (6.6)	0.006
During the past 7 days, I spent 2 or more hours doing chores (like mopping, vacuuming, sweeping, laundry, walking the dog, babysitting, yard work or other similar tasks at home or for others)	18 (32.7)	56 (61.5)	0.0007
During the past 7 days, I spent 2 or more hours doing extra physical activity (not counting what you do in school or other scheduled activities)	28 (50.9)	23 (25.3)	0.002
Sedentary activity
Screen time—weekday (total hours of screen time on weekdays reported, median (lower quartile, upper quartile))	16 (13, 20)	14 (11, 16)	0.005
Screen time—weekend (total hours of screen time on weekend reported, median (lower quartile, upper quartile))	16 (13, 20)	14 (12, 18)	0.007
(In my free time, on an average WEEKDAY, I spend 2 or more hours on…)
Playing video games	29 (52.7)	10 (11.0)	< 0.0001
(In my free time, on an average WEEKEND DAY, I spend 2 or more hours on…)
Playing video games	36 (65.5)	9 (9.9)	< 0.0001
Self-efficacy physical activity
Physical activity self-efficacy scale (median (lower quartile, upper quartile)) (possible range [5–25], higher score indicates higher self-efficacy)	22 (19, 27)	20 (15, 24)	0.002
I can exercise when feel bad about my body (4 or 5 on 5-point Likert scale)	48 (87.3)	53 (58.2)	0.0002
I can participate in a new physical activity (4 or 5 on 5-point Likert scale)	42 (76.4)	42 (46.2)	0.0003
I can be active when stressed (4 or 5 on 5-point Likert scale)	30 (54.6)	30 (33.0)	0.010
I can participate in a vigorous physical activity (4 or 5 on 5-point Likert scale)	37 (67.3)	41 (45.1)	0.009
Perceived barriers: physical activity
Perceived barriers scale: physical activity (median (lower quartile, upper quartile)) (possible range [15–60], higher score indicates more perceived barriers to physical activity)	30 (25, 34)	34 (29, 37)	0.0007
I get embarrassed if other kids see me being physically active (Agree/Strongly Agree)	5 (9.1)	32 (35.2)	0.0004
It’s too hard to exercise (Agree/Strongly Agree)	4 (7.3)	18 (19.8)	0.055
I am not motivated to exercise (Agree/Strongly Agree)	8 (14.6)	27 (29.7)	0.038
I don’t feel like exercising when in a bad mood (Agree/Strongly Agree)	20 (36.4)	49 (53.9)	0.040
Motivations for physical activity
Perceived benefits scale: physical activity (median (lower quartile, upper quartile)) (possible range [5–20], higher score indicates more perceived benefits of physical activity)	15 (15, 18)	15 (14, 16)	0.060
I have more energy when participating in regular physical activity (Agree/Strongly Agree)	50 (92.6)	73 (80.2)	0.045
Peer support healthy behaviors (Somewhat/very much)
Many of my friends are physically active	43 (79.6)	48 (52.8)	0.001
Social pressure from peers (Somewhat/very much)
Many of my friends influence me to be less physically active	1 (1.9)	11 (12.1)	0.030
Family influences on behaviors (Somewhat/very much)
My family members make healthy food choices	46 (85.2)	63 (69.2)	0.032
Self-esteem
Self-esteem scale (median (lower quartile, upper quartile)) (possible range [6–24], higher score indicates higher self-esteem)	17.5 (15, 19)	16 (15, 19)	0.068
At times I think I am no good at all (Agree/Strongly Agree)	17 (31.5)	51 (56.0)	0.004
Depression
Depression scale (median (lower quartile, upper quartile)) (possible range [6–18], higher score indicates higher risk of depression)	10 (8, 12)	11 (9, 14)	0.006
Body image
Body satisfaction scale (median (lower quartile, upper quartile)) (possible range [10–60], higher score indicates better body image)	38 (30, 44)	34 (30, 41)	0.167
I compare my body to the bodies of on magazines/social media
(I am satisfied with my…) (Agree of strongly agree)	5 (9.3)	32 (35.2)	0.0005
Stomach	19 (35.2)	14 (15.4)	0.006
Arms	37 (69.8)	36 (39.6)	0.0005
Emotional eating/eating triggers
Emotional eating scale (median (lower quartile, upper quartile)) (possible range [5–20], higher score indicates higher risk of emotional eating)	7 (6, 8)	8 (6, 10)	0.020
Weight control behaviors
I have been very or extremely successful in losing weight in the past	22 (40.7)	18 (19.8)	0.037
I have exercised in order to lose weight or keep from gaining weighting during the past month	50 (92.6)	68 (74.7)	0.008
I have paid attention to portion sizes in order to lose weight or keep from gaining weight during the past month	25 (46.3)	60 (65.9)	0.020

Questions were edited to fit the table format without changing any meanings

^*Chi-square test was used for categorical variables. Fisher’s exact test was used when the sample size was small, and Wilcoxon rank-sum test was used for continuous variables

Girls worried more about getting diabetes than boys (92.1% vs. 69.6%, p = 0.0004), and more boys than girls (58.9% vs. 33.0%, p = 0.002) reported that it would not be difficult to change diet and PA behaviors to prevent diabetes. While most diet-related behaviors and attitudes were similar between boys and girls, girls were more likely than boys to report that (1) they “eat unhealthily when in a bad mood” (53.9% vs. 30.4%, p = 0.005), (2) “unhealthy foods are too tempting” (68.1% vs. 41.1%, p = 0.001), and (3) they are “too rushed in the morning to eat healthy breakfast” (70.3% vs. 41.1%, p = 0.0005).

In terms of PA, boys reported an average of 6.6 h of moderate to vigorous PA every week, compared to 4.6 h in girls (p = 0.004), and more boys reported 2 + hours of strenuous exercise (60.0% in boys vs. 34.1% in girls, p = 0.002) and 20 + blocks per day of biking, skateboarding, scootering, and rollerblading (23.6% in boys vs. 6.6% in girls, p = 0.006). Boys also had a higher overall PA self-efficacy score (22, IQR (19, 27) vs. 20 (15, 24) in girls, p = 0.002) and were more confident in their ability to exercise when they feel bad about their body (87.3% in boys vs. 58.2% in girls, p = 0.0002). Boys were more likely to report that they would participate in a new PA (76.4% vs. 46.2% in girls, p = 0.0003). Boys also had fewer perceived barriers to PA (overall score 30 (25, 34) vs. 34 (29, 37) in girls, p = 0.0007), with more girls (35.2%) indicating that they “get embarrassed if other kids see them being physically active” than boys (9.1%, p = 0.0004). When asked about peer support and social pressure, boys were more likely to report that their friends are physically active (79.6% vs. 52.8% in girls, p = 0.001). While most PA behaviors and outcomes impacting physical activity were healthier in boys than girls, boys spent less time doing active chores (32.7% of boys vs. 61.5% of girls reported doing 2 or more hours of house chores in the past 7 days, p = 0.0007) and had more weekday and weekend sedentary time than girls. For example, more than half of the boys spent more than 2 h per day playing video games on weekdays and weekends (both p < 0.0001).

Girls had poorer self-esteem with 56.0% of girls agreeing with the statement “at times I think I am no good at all,” compared to 31.5% of boys (p = 0.004). Girls also scored higher on the depression scale (11 (9, 14) vs. 10 (8, 12) in boys, p = 0.006). In addition, girls were 3.5 times more likely to report that they compare their bodies to those in magazines and on social media (35.2% vs. 9.3% among boys, p = 0.0005). Girls were also less likely to report that they were satisfied with their stomach (15.4% vs. 35.3% in boys, p = 0.006) and arms (39.6% vs. 69.8% in boys, p = 0.0005). In addition, 92.6% of boys vs. 74.7% of girls reported exercising to lose weight/keep from gaining weight in the past month (p = 0.008).

Discussion

Summary

This study compared sex differences in measures related to metabolic health, lifestyle behaviors, and perceptions about diabetes and weight management among urban, racially and ethnically minoritized adolescents. We found a few notable differences between boys and girls. Girls were more likely to report that they were worried about getting diabetes and that they were concerned about getting health problems related to diabetes; however, boys (1) had fewer perceived difficulties in changing habits to prevent diabetes, (2) were more confident than girls in their ability to engage in certain healthy diet and PA behaviors, and (3) were more physically active (except for time spent doing physically active chores). Boys also reported fewer perceived barriers to healthy eating and PA, more social support for healthy eating and PA, better self and body image, fewer symptoms of depression, less emotional eating, and more success with previous weight loss attempts. Despite these protective factors, boys had more sedentary time, were more likely to engage in certain unhealthy dietary practices, and had higher overall cardiometabolic risk than girls. Our participants had a higher prevalence of prediabetes and metabolic risk than the general population, which was not surprising since overweight/obese individuals are more likely to develop these conditions [26]. These results increase our understanding of contributors to metabolic risk differences between boys and girls.

Clinical Outcomes

Clinical outcomes in our study often exceeded the upper limit of published normal ranges, which is not surprising as the parent clinical trial was designed to recruit racially and ethnically minoritized adolescents who were overweight/obese and at high risk for metabolic disorders such as prediabetes. We observed differences in clinical outcomes between boys and girls that align with previous studies; for example, body fat percentage is higher in girls than boys, regardless of weight status [27, 28]. The higher 2-h glucose and insulin levels in girls than boys are also consistent with previous findings [29, 30].

Compared to girls, boys in our study had a higher MetS score. The MetS score is a useful clinical and research construct for identifying individuals at risk for central obesity, insulin resistance, glucose intolerance, dyslipidemia, hypertension, other cardiovascular diseases, and other chronic illnesses [31, 32]. In youth, an elevated MetS score and subsequent diagnosis of metabolic syndrome or related conditions can be a valuable catalyst for intensive diet and exercise interventions to prevent further disease progression [33]. In our study, the higher average MetS score in boys than girls may indicate a greater risk of developing type 2 diabetes. However, the cutoff of MetS score to identify individuals at increased risk is unclear, and some literature suggests that the MetS score cannot be interpreted in the same way across studies, as every study has its own baseline population characteristics (e.g., geographic region, sex, age, race/ethnicity, socioeconomic status, and other clinical factors) [34]. Thus, the interpretation of the MetS score in specific populations may require additional evidence from longitudinal studies with these populations.

Diet and Physical Activity Behaviors, Attitudes, and Perceptions

We found differences between boys and girls in diet-related attitudes and barriers to a healthy diet. Boys also reported higher self-efficacy for certain healthy diet behaviors. One likely explanation is the overall confidence gap between boys and girls, which generally shows that males are more likely to be overconfident, while females are more likely to be underconfident regarding skills and performance of all kinds [35]. The confidence gap has been shown to impact every aspect of life, including both physical and mental domains, which may help explain why boys were more confident than girls in their perceived ability to engage in healthy behaviors and reported fewer perceived difficulties in changing habits to prevent diabetes. However, despite having higher healthy diet self-efficacy, boys were more likely to engage in certain unhealthy eating behaviors, which aligns with previous studies [36].

On average, girls reported less PA than boys, which is consistent across multiple studies, including a meta-analysis conducted on a global scale [37–39]. In addition, we found that boys engaged in more moderate to vigorous PA and strenuous exercise, while girls engaged in more physically active house chores. These findings may be explained by culturally prescribed gender roles [40]. A previous study found that doing house chores was a barrier to having an active lifestyle but did not examine differences by sex [41]. Known factors impacting differences in PA levels by sex include differences in opportunities to engage in extracurricular sports and encouragement for PA from family and friends [42], the latter of which is consistent with our findings.

Peer influence seems to have a different impact on diet and PA behaviors in girls and boys [43]. In addition, social influences from family members and friends impact adolescent PA and diet behaviors, and the effects are even stronger in girls [44]. This aligns with our findings that boys were more likely to be physically active because of peer support and eat healthier with their families, while girls were more likely to limit their PA and focus on dieting for weight loss due to peer influences. In terms of specific weight control behaviors, girls were more likely to include dietary strategies such as portion control, while boys focused more on exercise. These findings point to potential tailored strategies for behavioral weight management interventions based on sex.

We also found that girls reported more depressive symptoms and emotional eating and had a poorer self-image, body image, and body satisfaction than boys, which aligns with findings from previous studies [45–47]. Starting as early as 9–13 years, girls show more concern about being overweight than boys in the same age group [48]. Adolescent girls are also more susceptible to stress than boys, and girls are generally more self-conscious about weight and body image and display a higher level of worry and concern about weight-related conditions such as diabetes [49].

Study Limitations

Our study had some important limitations including its exploratory nature, as we only performed bivariate comparisons. Due to the small sample size, we were unable to conduct more advanced multivariable analyses to fully investigate the complex relationships between behaviors, behavioral determinants, and clinical outcomes. As a result, we must exercise prudence in our interpretation of the results and recognize that they are essentially exploratory and intended to generate hypotheses rather than provide definitive conclusions. Our study’s findings are limited to what can be inferred from basic comparisons, and it is not possible to draw conclusive associations without additional research. Moreover, we did not collect information about gender identity or sexual orientation and thus could not examine the intersectional effects of these identities and biological sex on eating and physical activity behaviors. Missing information about gender identity might also lead to gender misclassification and promote stereotypes with findings based solely on biological sex. In addition, we had a relatively small convenience sample from one community, which limits generalizability. Future studies should further explore sex and gender differences in metabolic risk with larger and more diverse populations across races/ethnicities, genders, ages, and geographic regions. In addition, many of our study findings are based on self-reported survey data, which could be subjected to response and recall bias. While acknowledging these limitations, it is noteworthy that our study is the first of its kind to comprehensively investigate these outcomes among racially/ethnically diverse urban adolescents residing in a community with a high prevalence of diabetes. As a result, our findings provide valuable descriptive information that may be pertinent to similar communities.

Conclusions

Our study provides initial insights into potential sex-based differences in dietary and PA behaviors and metabolic risk among urban Hispanic and Black adolescents, which may inform how future adolescent lifestyle interventions may be modified or tailored based on sex. For example, boys may have suggestions for increasing confidence and PA levels, while girls might contribute more with content related to nutrition label reading, portion control, reducing sedentary behaviors, and navigating healthy lifestyle challenges including social influences and difficult emotions. Future studies should focus on examining how lifestyle and disease prevention interventions impact behavioral and clinical outcomes based on sex and gender among adolescents.