General or Central Obesity and Mortality Among US Hispanic and Latino Adults

Key Points Question What are the associations of obesity with mortality among US Hispanic and Latino adults? Findings In this population-based cohort study of 15 773 US Hispanic or Latino adults with diverse backgrounds, greater waist to hip ratio was associated with higher mortality regardless of baseline body mass index and comorbidities, whereas severe obesity was only associated with higher mortality among those with unhealthy waist to hip ratio. Sex differences in the associations of body mass index and waist to hip ratio with mortality were observed. Meaning These findings suggest that prioritizing clinical screening and intervention of waist to hip ratio may be an important public health strategy among US Hispanic or Latino adults; however, sex-specific strategies might be needed.


Study Population
The HCHS/SOL is an ongoing population-based cohort study.At baseline (2008)(2009)(2010)(2011), four field centers from four metropolitan areas in the US -Chicago (Illinois), Miami (Florida), Bronx (New York), and San Diego (California) planned to recruit up to 4000 Hispanic/Latino adults (~1500 adults aged 18-44 and ~2500 adults aged 45-74) via a multi-stage probability sampling method.Finally, 16,415 Hispanic/Latino adults aged 18-74 years were recruited.Stable communities were selected so that participants could be contacted and examined over time.
The following individuals will be excluded: (1) those planning to move away in the next 3 years, (2) those with severe health problems, disabilities, or mental problems and unable to complete informed consent and study.Participants could choose to use English or Spanish, as the HCHS/SOL had both English and Spanish interviewers and questionnaires.Participants have diverse backgrounds including Central and South American, Cuban, Dominican, Mexican, and Puerto Rican.

Assessment of Obesity Parameters
Participants were asked to wear scrub suits or light clothing.BMI was calculated as body weight (kg) divided by squared height (m).Height (nearest cm) was measured using the wall-mounted Seca 222 stadiometer (Seca, Hamburg, Germany).Participants were asked to remove their hair ornaments from the top of the head.The back of the head, shoulder blades, buttocks, and heels contacted the vertical backboard.Weight (nearest 0.1 kg) and body fat percentage (BF%, nearest 0.1%) were measured using the Tanita Body Composition Analyzer TBF-300A (Tanita Corporation of America, Inc, Arlington Heights, IL, USA).BF% was estimated by the bioelectrical impedance method and corrected based on the 18O dilution method (corrected BF%=18.86+0.66*BF%for women; corrected BF%=12.75+0.73*BF%for men).BF% was categorized according to a previous HCHS/SOL study based on sex, age, and Hispanic/Latino background.In brief, sex-, age-, and Hispanic/Latino-backgroundspecific linear regression models for BF% regressed on 1/BMI were fitted, which were used to predict BF% corresponding with WHO BMI cutoffs, and cutoffs for lowest (corresponding to normal weight), second lowest (overweight), second highest (moderate obesity), and highest (severe obesity) BF% groups are shown in eTable 1. Waist circumference (WC) and hip circumference (both nearest cm) were measured using Gulick II anthropometric tape.The participants were asked to stand and hold their clothing above the waist.The pants and underclothing of the participant were slightly lowered.From behind and to the right of the participant, the staff palpated the hip area to locate the right ilium, drew a horizontal line just above the uppermost lateral border of the right ilium, and crossed the line to indicate the mid-axillary line of the body.Then, the staff stood on the participant's right side and placed the anthropometric tape around the trunk in a horizontal plane at the level marked on the right side of the trunk.The mirror on the wall was used to ensure the correct horizontal alignment of the anthropometric tape.The recorder (if available) ensured that the tape was parallel to the floor, snug, and not compressing the skin.Measurements were made at the end of a normal expiration.The hip circumference was measured at the level of maximal protrusion of the gluteal muscles, which was verified by passing the tape above and below the observed maximum.The anthropometric tape would be kept horizontally at this level, snug, but not compressing the skin, and the measurement should be made from the participant's right side.The mirror on the wall was used to ensure the correct horizontal alignment of the anthropometric tape.The recorder (if available) verified the position of the tape from both the front and back.WHR was calculated as WC divided by hip circumference.

Assessment of Covariates
All covariate information was self-reported and collected via questionnaires.Hispanic/Latino backgrounds included Central American, Cuban, Dominican, Mexican, Puerto Rican, South American, and more than one/others.Marital status was grouped into single, married/living with a partner, and separated/divorced/widow(er).Annual household income was grouped into less than $30,000 and $30,000 or more.Education was evaluated according to total years of schooling and highest grade/level of achieved education, which was categorized into no high school diploma/General Educational Development (GED; ≤10 years of education), at most a high school diploma/GED (11-12 years of education), and greater than high school/GED (≥13 years of education; highest level of trade/vocational school, university/college; or trade school certificate, associate degree, bachelor's degree, etc.).employment status was categorized into retired and not currently employed, not retired and not currently employed, employed part-time (≤35 hours/week), and employed full-time (>35 hours/week).Health insurance coverage was categorized into no and yes, and insurance provided through employers or labor unions, individual plan, Medicaid, Medicare, military insurance, Indian Health Services, etc. were considered.Participants were asked about their places of birth and years lived in the US, and accordingly, participants were categorized into living in the US 50 states/DC for less than 10 years, 10 years or more, and born in the US.Language preference was categorized into Spanish and English.Cigarette smoking status, exposure years, and the daily number of used cigarettes were asked.For current smokers, cigarette pack-years were calculated as the average daily number of used cigarettes multiplied by smoking years divided by 20 (number of cigarettes per pack), and the median of cigarette pack years was 6.30.Participants were categorized into never smoking, former smoking, current light smoking (current smoking and below the median of cigarette pack years), and current heavy smoking (current smoking and above the median of cigarette pack years).Participants were asked about their current alcohol drinking status and total weekly alcohol consumption, and alcohol use was categorized into never drinking, former drinking, current low to moderate drinking (<7/14 drinks/wk for women/men), and current heavy drinking (≥7/14 drinks/wk for women/men).Sleep duration was calculated according to weekday and weekend sleep duration and categorized into 6-9 h/d and <6/>9 h/d.Healthy Eating Index 2010 was calculated according to consumption of fruit, juice, vegetables, greens and beans, whole grains, dairy, total protein foods, seafood and plant protein, fatty acids, refined grains, sodium, and total energy, which was categorized into quintiles.Global physical activity questionnaire was used for physical activity measurements, and total physical activity (weekly moderate/vigorous physical activity) was classified into high activity (>300/150 min), medium activity (150-300/75-150 min), low activity (0-149/0-74 min), and inactive (no activity).Family history of myocardial infarction and diabetes, and prevalent diabetes, cardiovascular disease, cancer except for nonmelanoma skin cancer, and chronic obstructive pulmonary disease were self-reported by participants.Additionally, if participants used glucose-lowering medications or had fasting blood glucose level of ≥126 mg/dL, blood glucose level of 2-hour post-oral glucose tolerance test of ≥200 mg/dl, or glycosylated hemoglobin level of ≥6.5%, they would also be defined as diabetes cases.Cardiovascular disease included prevalent heart disease (i.e., self-reported physician-diagnosed heart attack, self-reported procedures of a balloon angioplasty, a stent, or bypass surgery, or electrocardiogram reports of major Q wave abnormalities or minor Q/QS waves with ST/T abnormalities) and cerebrovascular disease or carotid revascularization (i.e., self-reported stroke, mini-stroke or transient ischemic attack, or procedures of a balloon angioplasty or surgery to the arteries of necks to prevent or correct a stroke).Emphysema and chronic bronchitis were also considered as chronic obstructive pulmonary disease.

Subgroup Analyses
Subgroup analyses by sex, age, Hispanic/Latino background, acculturation, socioeconomic status, cigarette smoking, physical activity, and dietary quality were conducted.Multiplicative and additive interactions were evaluated by introducing a product term of the obesity parameter and subgroup variable into Cox proportional hazard regression and additive hazards model, and we focused on the interactions of the highest vs lowest categories of obesity parameters, which was of public health interest.Each subgroup variable was categorized into two levels.Sex was grouped into men and women.Age was grouped into <60 years and ≥60 years.Three major Hispanic/Latino backgrounds were considered, i.e., Cuban (yes and no), Mexican (yes and no), and Puerto Rican (yes and no).Acculturation was evaluated by an acculturation score according to the Multi-Ethnic Study of Atherosclerosis.A 5-point acculturation score was calculated according to nativity, years in the US, and language preference, with a higher score indicating more accultured.Nativity and years in the U.S. were combined to assign a score of 3 for US-born participants, 2 for foreign-born participants who lived in the US for over 20 years, 1 for foreign-born participants who lived in the US for 10-19 years, and 0 for foreign-born participants who lived in the US for less than 10 years.Preferred language was used to assign a score of 2 for preference to English, 1 for equal preference to English and Spanish, and 0 for preference to Spanish.The acculturation score was obtained by summing these scores, and participants were grouped into more accultured (scores of 3-5) and less accultured (scores of 0-2) according to the midpoint of the score range.A socioeconomic score was constructed according to income, education, and employment.Each socioeconomic factor was categorized into two groups; the advantaged group (i.e., yearly household income ≥$30,000, education level of high school or higher, and being employed) received one point, and the disadvantaged group received zero points.The sum of the points was the socioeconomic score, and participants were grouped into low (0-1 point) and high socioeconomic groups (2-3 points).Physical activity was grouped into high activity (weekly moderate/vigorous physical activity >300/150 min) and no high activity (≤300/150 min).Dietary quality was grouped into above and below the median of the Healthy Eating Index 2010.

More Stringent Criteria to Ascertain Death
In the main analysis, deaths were determined via death certificates, the National Death Index, and proxy reports during annual telephone follow-ups.All death records from death certificates and the National Death Index provided specific dates of death, which have been updated until December 31, 2019.Most proxy reports during annual telephone follow-ups provided specific dates of death, while dates of death were unavailable for 104 out of 686 deaths, and the date of the telephone follow-up will be used as the date of death.In the sensitivity analysis, we used more stringent criteria to ascertain deaths.This definition only included deaths identified via death certificates, the National Death Index, and proxy reports during annual telephone follow-ups that reported dates of death.Additionally, follow-up time started from baseline until the date of death, loss to follow-up, or December 31, 2019, whichever occurred first.Analyses considered complex survey design, including stratification, clustering, and sampling weights.Age, sex, field center, Hispanic background, marital status, income, education, employment status, insurance, length of staying in the US, preferred language, family history of myocardial infarction and diabetes, cigarette smoking, alcohol drinking, sleeping time, diet, and physical activity.
All P for multiplicative interaction>.05,except the association of waist-to-hip ratio between socioeconomic score (P=.031).CI indicates confidence interval; HR, hazard ratio.

eFigure 1 .
Flow Chart of the Study16,415 participants from HCHS/SOL(2008)(2009)(2010)(2011) 432 were excluded due to missing information on body fat parameters 71 missing information on body mass index 32 missing information on waist circumference 7 missing information on waist-to-hip ratio 322 missing information on body fat percentage 15,983 participants with complete information on body fat parameters 125 were excluded due to body mass index of <18.5 kg/m 2 85 were excluded due to death within the first two years of follow-up 15,773 participants were included in the analysis We used subpopulation or domain statements to restrict the analyses to the eligible participants, and those "excluded" participants also contributed to the variance calculations.HCHS/SOL indicates Hispanic Community Health Study/Study of Latinos.

eTable 1. Corrected Body Fat Percentage Corresponding to Body Mass Index Cutoffs by Sex, Age, and Hispanic/Latino Background a
a Values are corrected body fat percentage corresponding to the body mass index from a previous Hispanic Community Health Study/Study of Latinos.eTable 2.

Associations between Different Obesity Parameters and Mortality a
Analyses considered complex survey design, including stratification, clustering, and sampling weights.Deaths occurring within the first two years of follow-up were censored to reduce the possibility of reverse causation.Model 1 adjusted for age, sex, field center, Hispanic/Latino background, marital status, income, education, employment status, insurance, length of staying in the US, preferred language, family history of myocardial infarction and diabetes, cigarette smoking, alcohol drinking, sleeping time, diet, and physical activity.Model 2 additionally adjusted for prevalent diabetes, cardiovascular disease, cancer, and chronic obstructive pulmonary disease.
a The analysis considered complex and multistage probability sampling (i.e., sampling weight, stratification, and clustering), and all estimates were standardized to the United States 2010 Decennial Census age distribution.Data are shown as mean (standard error) for continuous variables and percentage (standard error) for categorical variables.eTable 3. a eTable 4.

Hazard Ratios (95% Confidence Intervals) Related to Different Obesity Parameters for Mortality: Sensitivity Analyses a
Analyses considered complex survey design, including stratification, clustering, and sampling weights.Age, sex, field center, Hispanic/Latino background, marital status, income, education, employment status, insurance, length of staying in the US, preferred language, family history of myocardial infarction and diabetes, cigarette smoking, alcohol drinking, sleeping time, diet, and physical activity were controlled.Analysis 1 excluded participants with prevalent diabetes, cardiovascular disease, cancer, and chronic obstructive pulmonary disease.Analysis 2 did not censor deaths occurring within the first two years of follow-up.Analysis 3 used a more stringent criteria to ascertain deaths.Analysis 4 only included never smokers.Analysis 5 used multiple imputations to impute missing covariates.Analysis 6 treated obesity parameters as time-varying variables by including the measurements at the first follow-up (2014-2017).Analysis 7 excluded participants from one field center at a time.b Groups of body fat percentage are defined in Supplementary a

Table 1. eTable 5. Distribution of Risk Factors by Sex and Obesity Parameters a
The analysis considered complex and multistage probability sampling (i.e., sampling weight, stratification, and clustering), and all estimates were standardized to the United States 2010 Decennial Census age distribution.Data are shown as mean (standard error) for continuous variables and percentage (standard error) for categorical variables.b Cigarette pack years were calculated as the average daily number of used cigarettes multiply smoking years divided by 20 (number of cigarettes per pack). a

Distribution of Risk Factors by Body Mass Index and Waist-To-Hip Ratio a
The analysis considered complex and multistage probability sampling (i.e., sampling weight, stratification, and clustering), and all estimates were standardized to the United States 2010 Decennial Census age distribution.Data are shown as mean (standard error) for continuous variables and percentage (standard error) for categorical variables.b Cigarette pack years were calculated as the average daily number of used cigarettes multiply smoking years divided by 20 (number of cigarettes per pack).
© 2023 Zhang Y al. JAMA Network Open. a