Language,Duration of United States Residency,and Leisure Time Physical Activity Among Women from the Third National Health and Nutrition Examination Survey (NHANES III)

Abstract

Background and Purpose:

Physical inactivity is a leading public health concern, particularly among women and ethnic minority groups, where Latinas are among the largest and fastest growing U.S. populations. Acculturation, known to affect other health behaviors, may explain low physical activity (PA) among these underserved women. Research on the effects of acculturation on PA, however, is scarce or limited by methodology. The study purpose was to evaluate the association between acculturation (i.e., language, birth country, and duration of U.S. residency) and PA in a national sample of women within the framework of the socioecologic model of health promotion.

Methods:

A total of 5,861 women (86% white, mean age 37.2) were sampled from the Third National Health and Nutrition Examination Survey (NHANES III).

Results:

Hierarchical multivariable regression modeling results indicate significant associations among language, duration of U.S. residency, and age, after controlling for confounders (all p<0.05). Most women reported less than recommended PA.

Conclusions:

These findings indicate that age, duration of U.S. residency, and language are important to consider in combination when understanding women's PA, findings that have implications for future research, theory, and clinical practice (e.g., making available PA assessments in multiple languages additionally tailored on age and investigating sociopolitical factors unique to Latinas).

Introduction

The numerous health benefits of participation in regular physical activity (PA) have been long established, making physical inactivity among the leading preventable causes of death and disease among U.S. adults (e.g., heart disease, diabetes, and obesity).^1

–5 Most Americans, however, do not meet national aerobic guidelines recommending 150 minutes/week of moderate intensity PA.^3,5
–7 The repeat call to action by Healthy People further highlights the epidemic inactivity of U.S. adults, most notably, the disproportionately high physical inactivity burden among certain U.S. subpopulations, such as women, particularly ethnic minorities.⁸ Thirty-five percent of women compared to almost 30% of men are physically inactive.⁹ Latinas are among the largest and fastest growing ethnic minority population in the United States. As a group, Latinas encompass multiple racial ancestries, including black and white, and many are Spanish speaking.¹⁰ Almost half of Hispanic women (∼49%) are aerobically inactive compared to roughly 30% of white women.⁹ These trends suggest unhealthy levels of physical inactivity among Latinas; and given that Latinas encompass multiple races, these trends call for efforts to examine the unique and complicated correlates of physical inactivity for these underserved people.

Ecological models suggest that multiple levels of factors (e.g., individual, environmental/policy) are needed to change or influence PA health disparities.^11

–14 The socioecological model of health promotion in particular specifies five nested levels of determinants of women's PA, where outward level factors have increasingly broader ability to impact women's social and physical PA environments and, therefore, impact inward level PA factors (Table 1). Acculturation is a socioecological factor that has been receiving greater attention in PA research. It has been conceptualized as the social and psychological exchanges, such as values, beliefs, and attitudes, that occur as a result of continuous contact and interaction between individuals from different cultures.^15

–18 Thus, acculturation has the propensity to influence multiple levels of PA factors. For example, the country a woman was born in may dictate her individual level of health, the community and region of the country she lives in, and immigration policies that affect her, all of which may impact her PA.^{7,11,18

–21} Much of the acculturation PA research, however, has been largely devoted to examining individual-level influences of acculturation on women's PA, using birth country, duration of U.S. residency, and language proxy measures.¹⁹ Findings to date are inconclusive.²² Among Spanish-speaking Latinas, language, duration of U.S. residency, and birth country acculturation may have negative,^23,24 positive,^{13,19,25
–27} no effect,^28,29 or mixed effects³⁰ on PA.

Table 1.

The Socioecological Model of Health Promotion and Women's Physical Activity

Number	Level	Description	Factors most relevant to women's inactivity	Present study variables
1	Intrapersonal (Individual) ^b	Biologic, developmental and psychologic (knowledge, skills, attitudes) factors	Getting older, racial minoirty or ethnic minority background, poor health status, low socioeconomic status (e.g., low education, poverty), acculturation^7,9,12,13	Age, race, ethnicity, health status (percieved health status, comorbidity index score, overweight/obesity), socioeconomic (education, federal poverty income ratio, occupation/work status), acculturation (birth country, duration of U.S. residence, and language spoken at home)
2	Interpersonal	Family, friends, and other social networks	Inadequeate social support, including friend, family, and partner support (referents)	—^a
3	Insitutional or organizational	Companies, schools, health agencies, and health facilities	Worksite and church health promotion programs, child care facilities¹³	—^a
4	Community ^b	Relationships among organizations, institutions, and social networks within a defined area	Urban residence, geographic region^12,13	Urban residence, U.S. geographic region
5	Policy	National, state, and local laws	Zoning laws for walking and biking trails, immigration laws¹³	—^a

Variable not available in Third National Health and Nutrition Examination Survey (NHANES III).

Factors available in NHANES III and examined in the present study.

These mixed findings may be the result of multiple socioecological variables of acculturation that have been inadequately accounted for by PA research. For example, one study found that age interacts with language, such that PA is lower among older Spanish-speaking women.²⁸ Others have found similar interactions between age and acculturation for other health conditions.³¹ Geographic region of residence, such as living in urban areas or a particular region of the United States, also is associated with physical inactivity due to multiple interacting factors. We were able to locate no quantitative research examining acculturation and PA within the socioecological model of health promotion. Findings from a focus group-based study (N=25) of low socioeconomic middle-aged Mexican women attending church in San Diego, California, highlighted cultural norms and environmental barriers to PA.³²

The present study examined acculturation and PA among diverse women from the Third National Health and Nutrition Examination Survey (NHANES III) to better understand physical inactivity among underserved Latinas. Although the socioecological model of health promotion is not directly tested in the present study, it served as the conceptual framework used to formulate hypotheses; select potential PA and acculturation confounders; plan analyses; and recommend future directions. All variables were selected based on their availability in the NHANES III dataset. Given that duration data were not available, a proxy PA outcome variable was calculated based on national guidelines.³³ We hypothesized the following: (1) Birth country and PA: Women born in the United States will have significantly higher PA scores than women not born in the United States; (2) Duration of U.S. residency and PA: Women who have lived a larger percent of their lives in the United States will have significantly higher PA scores than women who have lived a smaller percentage of their lives in the United States; (3) Language spoken at home and PA: Women who speak mostly English at home will have significantly higher PA scores than women who do not speak mostly English at home.

Materials and Methods

NHANES III was conducted between 1988 and 1994 by the National Center for Health Statistics (NCHS) of the Centers for Disease Control and Prevention (CDC). The procedures are described in detail elsewhere.³⁴ In brief, NHANES III used a complex, stratified, multistage probability design with clustered sampling to collect health status estimates on the civilian noninstitutionalized population aged ≥2 months. NHANES III is the first in its series that oversampled blacks and Mexican Americans. NHANES III recruited 39,665 eligible participants from whom we sampled women aged 20–59 who completed the medical examination at the mobile examination center and the home interview, including the household adult questionnaire and PA module. Participants classified as other (n=224) or Mexican American of unknown race (n=1) were excluded from analyses due to small sample sizes and potential interpretation difficulties. A total of 5,861 women met inclusion criteria. Institutional Review Board approval for the protection of human subjects was obtained from the University of Alabama at Birmingham before accessing NHANES III data.

Definition of variables

NHANES III data^35,36 were collected via standardized questionnaires administered during home interview and medical examination by trained bilingual interviewers.

Physical activity

PA variables were calculated using the household adult questionnaire, which asked about frequency of participation in nine specified activities during the past month (walking, jogging/running, bicycling, swimming, aerobics/aerobic dance, other dance, calisthenics/floor exercise, gardening/yard work, weight lifting) and up to four other activities not previously mentioned.³⁶ Duration information was not collected in NHANES III. The NCHS assigned metabolic equivalents (METs) for all activities; moderate-to-vigorous PA (MVPA)×MET=4–6.³⁷ Leisure time PA (main outcome) was calculated as a continuous variable by summing the MET by frequency (FREQ) product for each PA reported by a participant according to Dowda et al.³³ To meet MVPA recommendations of 150 minutes/week, a woman would need a MET×FREQ PA score of 86.67–130 (minimum recommended MVPA=4 METs×21.67 monthly PA FREQ=86.67; maximum recommended MVPA=6 METs×21.67 Monthly PA FREQ=130 ([recommended weekly PA FREQ×number of weeks per year]/number of months in a year=[5×52)/12=21.67 for each activity]). Lower scores indicate less PA.

Acculturation

Acculturation factors (family questionnaire) in the present study were birth country (U.S., not U.S.), years of U.S. residency (continuous variable, only participants not U.S. born), and language spoken at home (English, Spanish/other [primarily Spanish, both Spanish and English, or another language]).

Demographics

Nine demographic covariates were analyzed based on previous literature.¹² Age, race (white, black), ethnicity (Mexican American, other Hispanic, not Hispanic), Census region (Northeast, Midwest, South, West), urban residence (yes, no), and number of household members were assessed by the household screener.³⁴ Poverty income ratio (≥poverty, below poverty) was computed using variables from the NHANES III family questionnaire,³⁴ including family income, poverty threshold, the age of family questionnaire completer, and year of interview (ratios below 1.00=federal poverty level).^38,44 Other socioeconomic variables were education (<high school, high school, some college or higher), marital status (married [currently married, separated, or domestic partnership/cohabitating], not married [single, widowed, or divorced]), and occupation (blue collar, white collar, homemaker, going to school, unemployed [retired, disabled, laid off, other]).

Health status

We analyzed three health status covariates: perceived health status (good to excellent, fair to poor), body mass index (BMI, continuous, calculated based on measured height and weight by NCHS),³⁹ and comorbidity. The Charlson Comorbidity Index^40
–42 measured comboridity based on household adult questionnaire variables. Participants indicated (yes, no) if they had been told by their doctor that they had the any of following selected conditions that we weighted numerically based on disease severity: (1) myocardial infarction, chronic heart failure, stroke, chronic obstructive pulmonary disease (COPD) (chronic bronchitis and emphysema), and connective tissue diagnosis (lupus or rheumatoid arthritis) and (2) any tumor (skin cancer or lupus) and diabetes. Comorbidity score was calculated by summing the total number of endorsed (weighted) conditions (range 0–12).

Analyses

Statistical analyses were conducted using SAS version 9.3 (SAS Institute, Cary, NC) following recommendations discussed in Gossett et al.⁴³ Analyses incorporated sampling weights and stratification variables to account for NHANES III design complexity.⁴⁴ Sampling and stratification variables corrected for differential probabilities of selection, noncoverage, and nonresponse. Chi-square (categorical variables) and t tests (continuous variables) were used to analyze participant acculturation and socioecological characteristics and PA patterns. Given that acculturation factors were central to hypothesis testing, we used standard statistical significance criteria to retain variables for further analyses. Acculturation factors that demonstrated a significant bivariate association with PA at p<0.05 were retained for higher-order multivariable analyses. Socioecological variables significantly associated with PA during bivariate analyses were further analyzed as potential confounders using chi-square and logistic regression (categorical outcome) and simple regression (continuous outcome). A variable was considered a confounder if it was statistically associated with both PA and an acculturation predictor (p<0.05). Multivariable analyses controlled for these confounders. Socioecologic variables that demonstrated a consistent association with women's' PA in previous research or were central to hypothesis testing were also retained during higher-order analyses.⁴⁵ First order multivariable models regressed PA on each acculturation×age interaction term and relevant main effects if an acculturation predictor demonstrated a significant association with PA during bivariate analyses. Final multivariable regression models tested the associations between PA and each acculturation variable, adjusting for significant acculturation×age interaction and main effect terms and socioecologic confounders from lower-order models. A natural log transformation was originally applied to the PA variable, given that it was positively skewed and violated assumptions of normality. Transforming the PA variable did not result in statistical findings that were different from findings using the nontransformed PA variable; thus, we have reported nontransformed PA findings for ease of results interpretation.

Results

Participant characteristics

Descriptive statistics for the aggregated sample are presented in Table 2. Almost 89% of women reported that they were born in the United States, with the average reported length of U.S. residency at 14.5 years. About 92% of women spoke English, almost 6% spoke Spanish, and the remaining 1.3% spoke both English and Spanish at home. Over 59% of participants were<39 years of age, with a mean age for the sample of approximately 37 years. The sample was 86% white and 90% non-Hispanic. About 75% of women reported being married, 44% reported completing≥some college, 43% were homemakers, and 14% lived below the federal poverty level. Just over one third of women reported living in the South, and almost one quarter reported living in the Midwest. A total of 48% of women reported living in urban areas. Almost 81% of women reported no comorbid conditions, and 50% were classified as overweight or obese. Roughly 87% reported good or better health.

Table 2.

Participant Acculturation, Socioecological Characteristics, and Physical Activity Scores (N=5,861)

	n^a	Prevalence ^b % (SE)	Physical activity score ^c Mean (SE)	PA score p^d
Birth country
U.S.	4,637	88.63 (1.02)	11.14 (0.61)	0.06
Not U.S.	1,224	11.37 (1.03)	6.77 (1.42)
Duration of U.S. residency, years
< 5	267	18.25 (2.43)	2.45 (0.67)	<0.01^*
5–9	226	21.59 (3.05)	2.67 (0.80)
10–19	357	27.66 (2.08)	11.08 (3.45)
> 20	220	32.50 (3.19)	8.34 (3.48)
Language spoken at home
English	4,668	92.38 (0.84)	11.33 (0.62)	<0.01^*
Not primarily English^e	1,193	7.62 (0.84)	2.31 (0.37)
Age, years
20–29	1,795	28.41 (1.15)	9.50 (0.77)	0.03^*
30–39	1,785	31.00 (1.13)	7.59 (0.71)
40–49	1,314	24.25 (0.95)	7.42 (0.65)
50–59	967	16.34 (0.68)	5.72 (0.78)
Race
Black	2,091	13.83 (0.93)	6.17 (0.66)	<0.01^*
White	3,770	86.17 (0.93)	8.75 (0.56)
Ethnicity
Hispanic	1,808	9.88 (0.96)	5.31 (0.64)	<0.01^*
Not Hispanic	4,053	90.12 (0.96)	8.95 (0.50)
Marital status
Married	4,126	76.99 (1.08)	7.75 (0.45)	0.71
Not married	1,721	23.01 (1.08)	8.03 (0.81)
Education
< High school	1,852	18.93 (1.22)	3.84 (0.63)	<0.01^*
High school	2,141	37.48 (1.10)	7.04 (0.61)
Some college or higher	1,868	43.59 (1.49)	12.68 (0.72)
Occupation
Blue collar	332	8.71 (0.79)	6.35 (1.16)	0.32
White collar	507	20.99 (1.33)	11.49 (1.87)
Homemaker	1,649	42.92 (1.53)	5.75 (0.72)
Attending school	233	6.99 (0.69)	10.25 (2.40)
Unemployed^f	586	20.39 (1.20)	8.17 (1.14)
Live below poverty
Yes	1,931	18.92 (0.99)	5.24 (0.66)	<0.01^*
No	3,930	85.76 (0.99)	9.10 (0.50)
Census region
South	2,551	35.86 (2.66)	6.37 (0.63)
Northeast	796	20.68 (1.29)	9.47 (1.27)	<0.01^*
Midwest	1,126	23.55 (1.09)	10.12 (0.52)
West	1,388	19.91 (3.62)	7.72 (1.04)
Urban residence
Yes	3,032	48.47 (4.92)	7.93 (0.56)	0.08
No	2,829	51.53 (4.92)	7.73 (0.72)
Comorbidity score
No conditions	4,781	80.94 (0.71)	7.96 (0.46)	0.35
≥ 1 conditions	1,080	19.05 (0.71)	7.25 (0.99)
Body mass index
Under weight or normal weight	2,417	51.45 (1.36)	10.31 (0.67)	<0.01^*
Overweight	1,627	24.95 (0.90)	7.23 (0.66)
Obese	1,817	25.46 (1.21)	5.06 (0.57)
Perceived health status
Good or better	4,604	86.82 (0.83)	8.58 (0.47)	<0.01^*
Fair to poor	1,257	13.18 (0.83)	5.09 (0.79)
Total	5,861	–	10.64 (0.59)	–

n, nonweighted sizes of actual subsamples in the present study.

Prevalence, weighted estimates of population parameters—%, standard error (SE)—based on present study subsamples.

Physical activity (PA) score, average (per participant) sum of the frequency of participation (monthly) by metabolic equivalent (MET) products for each reported activity by a participant; a score within the range of 86.67–130 would be necessary to meet moderate to vigorous physical activity recommendations in the present study.

p-value represents the unadjusted statistical significance test for differences in PA score (outcome) across levels of each variable using simple regression.

Includes Spanish-speaking, both English and Spanish-speaking, and other language categories.

Denotes participants who reported being disabled, laid off, or retired.

Significant values, where p<0.05 for acculturation predictors and socioecological variables.

Physical activity patterns

The average PA score was 10.64 (standard error [SE] 0.59). A total of 62.28% of women reported no PA (SE 1.19, n=4,097); 34.23% of women reported participating in PA, but at less than recommended amounts (SE 1.15, n=1,618); and 3.49% reported engaging in recommended PA levels (SE 0.29, n=146). Walking for PA was reported by 53.97% of women (SE 1.16, n=2,760), with an average reported frequency of walking participation in the past 30 days of 1.52 (0.01).

Findings from unadjusted bivariate modeling of physical activity by acculturation and socioecological factors are presented in Table 2. Duration of US residency and language spoken at home were each significantly associated with women's reported leisure time PA. Women who lived in the United States for <5 years or 5–9 years had significantly lower PA scores than women who had lived in the United States for 10–19 years or >20 years. Speaking primarily Spanish, both Spanish and English, or another language at home vs. English was associated with significantly less PA for women. Birth country was not statistically associated with women's PA and, therefore, was dropped from further analyses (p=0.06). All socioecological variables demonstrated significant group differences in PA (p<0.05) with the exception of marital status, occupation, and comorbidity score, which were dropped from further analyses.

Acculturation characteristics

Bivariate analyses revealed that all variables except Census region and BMI were significantly associated with duration of U.S. residency (p<0.05; results not shown), including language, age, race, ethnicity, education, poverty income ratio, urban residence, and perceived health status. Bivariate analyses also found that the following variables were associated with language spoken at home (p<0.05; results not shown): age, race, ethnicity, education, poverty income ratio, BMI, perceived health status, geographic region, and urban residence.

Multivariable regression modeling

Duration of U.S. residency and PA

Results of first order multivariable modeling of the association between duration of U.S. residency, age, and duration of U.S. residency×age and PA revealed a significant omnibus test, F(3,46)=1304.10, p<0.01, where a significant duration by U.S. residency×age interaction was observed: F (1, 46)=1171.16, p<0.01. The effect of age on PA varied depending on the number of years of U.S. residency. To illustrate, the effect of age on PA was positive for women living in the United States for 5 years (β=0.11, SE=0.05, t(46)=2.20, p=0.03), not significant for women living in the United States for 6 years (β=− 0.08, SE=0.05, t(46)=1.71, p=0.09), and negative for women living in the United States for 13 years (β=−0.09, SE=0.05, t(46)=−2.05, p=0.046) and 20 years (β=−0.27, SE=0.04, t(46)=−6.58, p<0.01).

Table 3 lists results of the final model regressing PA on duration of U.S. residency and age main effects and the interaction term, adjusted for socioecological confounders. The duration of U.S. residency×age interaction continued to be significantly associated with women's PA. Post-hoc contrasts revealed that the effect of age on women's PA maintained statistical significance, only for women who reported living in the United States for≥13 years, and age was negatively associated with PA for these women: β_{13 years}=−0.14, SE=0.06, t(46)=−2.12, p=0.04; β_{15 years}=−0.19, SE=0.06, t(46)=−2.90, p<0.01; and β_{20 years}=−0.32, SE=0.07, t(46)=−4.84, p<0.01. Among socioecological variables, race, education, poverty status, urban residence, and perceived health status maintained statistically significant associations with women's PA.

Table 3.

Final Multivariable Regression Model Examining Duration of U.S. Residency, Language Spoken at Home, and Women's Physical Activity, Adjusted for Socioecological Confounders

	β (SE)	t	p
Duration of U.S. residency^a
Intercept	−12.38 (1.52)	−8.14	<0.01^*
Duration of U.S. residency	1.20 (0.06)	20.31	<0.01^*
Age	0.20 (0.06)	3.04	<0.01^*
US residency×age^b	−0.03 (0.0005)	−47.98	<0.01^*
Race
Black	—^c	—	—
White	6.18 (1.26)	4.92	<0.01^*
Ethnicity
Hispanic	—	—	—
Not Hispanic	11.72 (1.74)	6.72	<0.01^*
Education
<High school	—	—	—
High school	−7.97 (0.96)	−8.27	<0.01^*
Some college or higher	−3.31 (0.66)	−5.01	<0.01^*
Live below poverty
Yes	—	—	—
No	1.38 (0.79)	1.76	<0.01^*
Urban residence
Yes	—	—	—
No	1.19 (0.59)	2.01	0.05
Perceived health status
Good or better	—	—	—
Fair to poor	3.14 (0.77)	4.08	<0.01^*
Language^d
Intercept	8.32 (3.33)	2.50	0.02^*
Language spoken at home
English	—	—	—
Not English	−12.24 (3.49)	−3.51	<0.01^*
Age	−0.11 (0.07)	−1.54	0.13
Language×age^e	0.20 (0.08)	2.39	0.02^*
Race
Black	—	—	—
White	3.46 (1.05)	3.28	<0.01^*
Ethnicity
Hispanic	—	—	—
Not Hispanic	1.36 (1.31)	1.03	0.31
Education
< High school	—	—	—
High school	0.56 (1.36)	0.41	0.68
Some college or higher	7.10 (1.50)	4.73	<0.01^*
Census region
South	—	—	—
Northeast	0.33 (1.38)	0.24	0.81
Midwest	4.35 (1.09)	4.00	<0.01^*
West	0.87 (1.87)	0.47	0.64
Body mass index
Normal weight or underweight	—	—	—
Overweight	−2.57 (1.37)	−1.88	0.07
Obese	−5.28 (1.26)	−4.21	<0.01^*

n=1,070. Final model omnibus F (5, 46)=2407.26, p<0.01.

U.S. residency×age values=the effect of increasing age on length of residence in the United States. Omnibus interaction effect, F (1, 46)=2301.61, p<0.01.

Referent groups for categorical variables are denoted by dashes.

n=5,861. Final model omnibus F (12, 49)=53.88, p<0.01.

Language×age values=the effect of age when language is not English. Omnibus interaction effect F(1, 49)=5.72, p=0.02.

Significant values denoted by^*, where p<0.05. p values represent dummy-coded comparisons within each variable in multivariable regression analyses.

Language and PA

Results of the first order multivariable model, regressing PA on language, age, and the language×age interaction revealed a significant omnibus test, F (3, 49)=54.68, p<0.01. The language×age interaction was significantly associated with PA: F (1, 49)=4.60, p=0.04. The effect of age on PA was positive for non-English-speaking women (β=0.18, SE=0.08, t (49)=2.15, p=0.04) and not significant for English-speaking women (β=0.02, SE=0.04, t (49)=0.38, p=070).

Results of the final model regressing PA on language, age, and language×age, adjusting for socioecological confounders, are shown in Table 3. The language x age interaction continued to be significantly associated with women's PA. Post-hoc contrasts revealed that among non-English-speaking women, increasing age was associated with higher reported PA. Race, education, geographic region, and BMI maintained statistically significant associations with women's PA.

Discussion

Findings from our study using ecological theory to examine women's PA behavior and acculturation demonstrated that from1988 to 1994, even though over half of women reported walking as a form of leisure time PA, almost two thirds (62%) were completely inactive, a third (34%) reported engaging in any PA, and relatively few (3.5%) reported engaging in enough PA to meet current recommendations. Among Latinas in particular, PA levels in our study were substantially lower compared to non-Latinas. Although PA estimates are lower in our study, the general pattern is consistent with findings from other national surveys among Latinas. A sizable number of Mexican American women from another NHANES III study were found to be completely inactive (46%), and more recent findings from NHANES (1999–2002) showed that 42.3% of Mexican American women engaged in any PA.³⁰ The National Health Interview Survey shows that roughly half (48.6%) of Latinas vs. less than a third (29.9%) of white women were completely inactive.⁹ The low PA prevalence estimates for women in general and Spanish-speaking Latina women in particular in our study potentially suggest that substantial gains in women's PA behavior have occurred in the recent decades since 1994.^30,46 However, more work is needed to improve Latina PA, given they are still among the least physically active groups in the United States.^7,47

Duration of U.S. residency and PA

The discrepancies in findings across studies, including ours, regarding higher physical inactivity and lower numbers of women meeting PA recommendations may be explained by a number of factors. The broader sample of Latinas in our study and the NHIS study⁹ vs. Mexican American women only samples used in NHANES studies^19,30 and the use of different measures to define/calculate PA in the various studies could account for these discrepancies.^9,19,30 Also, PA guidelines have changed in recent years, and inflated trends in recent PA prevalence estimates have been documented as a consequence of the newer guidelines.⁶ The discrepancies may also be explained by the fact that these estimates are based largely on models that inadequately adjust for socioecological confounders. In fact, when we adjusted for these factors and examined acculturation in particular, we found a negative effect of age on PA for women who had lived in the United States for an extensive amount of time, >13 years. The negative correlation between age and PA is well understood.^{6,12,33,48

–51} The interaction between age and duration of U.S. residency is unique, however, and in contrast to the typically positive main effect of longer duration of U.S. residency on PA found by others.^19,26,47 This suggests that as an immigrant woman's time living in the United States becomes longer, beyond 13 years, her PA begins to approximate the low levels observed by natural born U.S. women, where getting older on average is associated with reduced PA. It is notable that the effect of age on duration of U.S. residency<13 years was not significant, supporting the notion that age and duration of U.S. residency are not necessarily correlates of one another but, in fact, are separate constructs that interact to affect Latina PA. They, therefore, should be accounted for in future research examining duration of U.S. residency and women's PA.

Language and PA

In contrast to duration of U.S. residency, we found a positive effect of age on language acculturation among non-English-speaking women, where PA was reportedly higher as age increased. These findings, given the comprehensive control of confounders, add robust support to the growing majority of literature that suggests that low language acculturation is associated with reduced PA.^{13,19,25,26,47} The positive effect of age on Latina PA in this study appears to be contrary to what would be expected, given the large body of research showing a negative correlation of age on PA.^{6,12,33,48

–51} We were able to locate only one prior study that examined the interaction effect of age and language acculturation on Latina PA. Cantero et al.²⁸ found that among a regional sample of middle-aged and elderly Latinas, the effect of low language acculturation (speaking Spanish vs. English) on women's PA was stronger among younger (46–64 years) vs. older Latinas (65–74 years). Our study represents the first to extend the effect of age on language acculturation and Latina PA to a national and representatively younger sample of women, aged 20–59, and with comprehensive control for socioecological confounders.

Our findings suggest that similar to Cantero et al.,²⁸ whereas low language acculturation is detrimental to participation in PA, getting older helps improve PA among Latinas. Age in this case may be serving as a proxy for increases in opportunities to be active over time as the result of becoming more fluent in English and, therefore, less isolated in the community, and attainment of higher socioeconomic status.^19,28 According to U.S. Census data, almost half of all U.S. immigrants are Spanish-speaking women not fluent in English. Many of these women have less than a high school education and live in largely homogeneous impoverished communities in major urban areas.¹⁰ Thus, the possibility for less isolation is further illustrated by our finding of a marginal effect of urban residence on lower PA in the language acculturation model and a negative association between education and PA. Increases in opportunity as a result of language acculturation may also be reflected by child-rearing responsibilities that may become less burdensome over time for young to middle-aged women as their children and grandchildren become older/more independent.²⁸ Parenting variables were not available in NHANES III; thus, we could not examine the effect of children and parenting on Latina acculturation and PA, an area that merits further exploration in future studies.

The socioecological model of health promotion

The present study did not directly test the socioecological model of health promotion, but it served as a framework to guide the present study using variables available in NHANES III, a cross-sectional study not designed specifically to test acculturation and PA. We examined acculturation and PA accounting for two levels of confounders from the socioecological model of health promotion among the total five in the model (intrapersonal, interpersonal, organizational, community, environmental/policy),¹⁴ and not other levels of ecological factors known to affect women's PA, such as family, friend, and worksite/church support.¹² The framework was useful nonetheless. Depending on the acculturation factor of interest, different socioecological variables were retained or had different effects in the model. For example, race and education (and age) were the only variables in common between the language and duration of U.S. residency models of PA. Further, the effect of education was reversed for the two models, where higher education was associated with lower PA when duration of U.S. residency was accounted for in the model, and vice-versa for language. The reverse effects of age on PA across duration of U.S. residency (negative) vs. language (positive) observed in our study are also noteworthy and are likely explained in part by different socioecological variables in the model. Further, there may be an interaction among age, duration of U.S. residency, and language spoken at home related to PA that we were not able to test in the present study due to small sample sizes, that warrants further examination.

Other investigators have inferred the economic consequences of being an immigrant implied by duration of U.S. residency and language barriers to leisure time PA.^19,52 It may be however that age, similar to duration of U.S. residency and language is a proxy measure of discrimination and socioeconomic status (beyond education and income) that result in inferior work and housing conditions and less educational opportunity (organizational or policy/environmental level socioecological factors). Discrimination and poor socioeconomic status may also result in isolation and loss of social networks/support (interpersonal and organizational level socioecological factors) and related stress (intrapersonal socioecological factors). These factors may be more relevant for younger vs. older Latinas because of coping response, greater child care responsibilities, and lack of access to PA resources.^26,48 In fact, the effect of policy and the sociocultural environment of any given period and coping response and resources would be relevant to PA participation among diverse women and likely cause cohort effects regarding PA participation. What remains to be investigated are the impacts of economic climate, political laws, and public perception on immigration and the effect of these factors on Latina PA then and now. Within the context of the socioecological model of health promotion, we can derive a theory-guided, valid, and parsimonious model of Latina PA that can be used to develop tailored and, therefore, more effective PA promotion interventions. PA materials directed at women based on their length of U.S. residency and informed by available research may benefit by offering suggestions specific to the age of the women and by normalizing and discussing/providing tips to mitigate, for example, the decreases in PA for women who have lived in the United States for long periods of time as they get older. Research is underway investigating the effect of providing Spanish language materials for PA promotion among Latinas as recommended,¹⁹ and findings are promising.^53,54 We are unaware, however, of any current Latina PA promotion studies that tailor language-based materials based on age, a benefit that would be indicated based on findings in our study (e.g., providing tips for managing competing responsibilities, including child care among younger vs. older Latinas).

Limitations

Although we believe the findings from this study extend the evidence regarding acculturation and women's PA, there are limitations. The present study generalizes to women who are black or white racially or Mexican American and who indicate that they speak English, Spanish, or some combination of the two languages. PA duration and specific intensity information were not available in NHANES III, which limits comparison of these findings to other studies similar in design. Data were largely limited to self-report vs. objective measures, which are known to have problems with accuracy and validity in some cases, particularly among low-income and low-education samples.^55,56 However, the fact that the overwhelming majority of study participants are largely physically inactive based on self-report suggests that the poor health implications of physical inactivity may be a larger issue than likely PA misreporting and lack of information on PA duration. Further, understanding a woman's perception of the amount of PA in which she believes she is engaging is still relevant for appropriate treatment development. The focus of the present study was on leisure time PA, a construct that may hold a different meaning for different groups of women, particularly Latinas, and, thus, may not accurately portray PA levels for this group.¹⁹ Indeed, some evidence suggests that when occupation and transportation activity are accounted for, PA is higher among Latina vs. white women.²⁵

Conclusions

The present study served to broaden our understanding of acculturation and Latina women's PA behavior, using a comprehensive, large, representative national sample and guided by ecological theory. We found significant associations between age and women's PA across duration of U.S. residency and language, suggesting a more complicated relationship between these factors and PA than illustrated by previous research. We are limited in our understanding of the full combined influences of age and acculturation and their association with women's PA because of information not available in the present dataset that warrants examination in future research, including the associations among parenting and child care, social support, sociopolitical context, discrimination, coping strategies, and occupation and transportation and PA. Latinas remain underserved and among the largest ethnic minority groups in the United States. These facts, together with our findings, call for the development of tailored PA promotion interventions for Latinas using ecological theory toward this aim, where age and acculturation together serve as combined tailoring factors vs. either factor alone, in contrast to findings from previous research.

Footnotes

Acknowledgments

This study was based on dissertation work completed in the department of psychology at the University of Alabama at Birmingham. We thank Drs. Joshua Klapow, Bonnie Sanderson, Rudolph Vuchinich, Stacey Kovac, Beth Bock, Nancy Felipe-Russo, and Shira Dunsiger for their assistance with this project.

Disclosure Statement

No competing financial interests exist for any author.

References

Mokdad

, Marks

, Stroup

, Gerberding

. Actual causes of death in the United States, 2000. JAMA, 2004; 291:1238–1245.

Heron

. Deaths: Leading causes for 2007. Hyattsville, MD: National Center for Health, 2011; 59:1–95.

Cory

, Ussery-Hall

, Griffin-Blake

et al. Prevalence of selected risk behaviors and chronic diseases and conditions—Steps communities, United States, 2006–2007. MMWR Surveill Summ, 2010; 59:1–37.

US Department of Health and Human Services. Physical activity and health: A report of the Surgeon General. Atlanta: USDHSS, Centers for Disease Control and Prevention, 1996.

Physical Activity Guidelines Advisory Committee. Physical activity guidelines advisory committee report, 2008. Washington, DC: U.S. Department of Health and Human Services, 2008.

Carlson

, Fulton

, Schoenborn

, Loustalot

. Trend and prevalence estimates based on the 2008 Physical Activity Guidelines for Americans. Am J Prev Med, 2010; 39:305–313.

Schoenborn

, Adams

. Health behaviors of adults: United States, 2005–2007. Vital Health Stat, 2010; 10:1–132.

US Department of Health and Human Services. Office of Disease Prevention and Health Promotion. Healthy people 2020. healthypeople.gov/. 2012 April 25.

Schiller

, LJ , Ward

, Peregoy

. Summary health statistics for U.S. adults: National Health Interview Survey, 2010. Vital Health Stat, 2012; 10,252.

10.

Frequently requested statistics on immigrants in the United States. Migration Policy Institute, 2009. www.migrationinformation.org/. 2011 April 20.

11.

Sallis

, Owen

. Ecological models. Glanz

, Lewis

, Rimer

. Health behavior and health education: Theory, research, and practice, 2nd. San Francisco, CA: Jossey-Bass, 1997; 403–424.

12.

Eyler

, Wilcox

, Matson-Koffman

et al. Correlates of physical activity among women from diverse racial/ethnic groups. J Womens Health Gend Based Med, 2002; 11:239–253.

13.

Abraido-Lanza

, Chao

, Florez

. Do healthy behaviors decline with greater acculturation? Implications for the Latino mortality paradox. Soc Sci Med, 2005; 61:1243–1255.

14.

McLeroy

, Bibeau

, Steckler

, Glanz

. An ecological perspective on health promotion programs. Health Educ Q, 1988; 15:351–377.

15.

Redfield

, Linton

, Herskovits

. Memorandum for the study of acculturation. Am Anthropologist, 1936; 38:149–152.

16.

Sabogal

, Marin

, Otero-Sabogal

. Hispanic familism and acculturation: What changes and what doesn't? Hispanic J Behav Sci, 1987; 9:397–412.

17.

Ryder

, Alden

, Paulhus

. Is acculturation unidimensional or bidimensional? A head to head comparison in the prediction of personality, self-identity, and adjustment. J Pers Soc Psychol, 2000; 79:77–88.

18.

Berry

. Immigration, acculturation, and adaptation. Appl Psychol, 1997; 46:5–33.

19.

Crespo

, Smit

, Carter-Pokras

, Andersen

. Acculturation and leisure-time physical inactivity in Mexican American adults: Results from NHANES III, 1988–1994. Am J Public Health, 2001; 91:1254–1257.

20.

Brownson

, Baker

, Housemann

, Brennan

, Bacak

. Environmental and policy determinants of physical activity in the United States. Am J Public Health, 2001; 91:1995–2003.

21.

World Health Organization. Physical inactivity: a global health problem, 2011. www.who.int/dietphysicalactivity/factsheet_inactivity/en/. 2011 April 20.

22.

Lara

, Gamboa

, Kahramanian

, Morales

, Bautista

. Acculturation and Latino health in the United States: A review of the literature and its sociopolitical context. Annu Rev Public Health, 2005; 26:367–397.

23.

Esparza

, Fox

, Harper

et al. Daily energy expenditure in Mexican and USA Pima indians: Low physical activity as a possible cause of obesity. Int J Obes Relat Metab Disord, 2000; 24:55–59.

24.

Stern

, Gonzalez

, Mitchell

, Villalpando

, Haffner

, Hazud

. Genetic and environmental determinants of type II diabetes in Mexico City and San Antonio. Diabetes, 1992; 41:484–492.

25.

Berrigan

, Dodd

, Troiano

, Reeve

, Ballard-Barbash

. Physical activity and acculturation among adult Hispanics in the United States. Res Q Exerc Sport, 2006; 77:147–157.

26.

Evenson

, Sarmiento

, Ayala

. Acculturation and physical activity among North Carolina Latina immigrants. Soc Sci Med, 2004; 59:2509–2522.

27.

Jurkowski

, Mosquera

, Ramos

. Selected cultural factors associated with physical activity among Latino women. Womens Health Issues, 2010; 20:219–226.

28.

Cantero

, Richardson

, Baezconde-Garbanati

, Marks

. The association between acculturation and health practices among middle-aged and elderly Latinas. Ethn Dis, 1999; 9:166–180.

29.

Banna

, Kaiser

, Drake

, Townsend

. Acculturation, physical activity and television viewing in Hispanic women: Findings from the 2005 California Women's Health Survey. Public Health Nutr, 2012; 15:198–207.

30.

Ham

, Yore

, Kruger

, Heath

, Moeti

. Physical activity patterns among Latinos in the United States: Putting the pieces together. Prev Chronic Dis, 2007; 4:A92.

31.

Jaber

, Brown

, Hammad

, Zhu

, Herman

. Lack of acculturation is a risk factor for diabetes in Arab immigrants in the U.S. Diabetes Care, 2003; 26:2010–2014.

32.

Martinez

, Arredondo

, Perez

, Baquero

. Individual, social, and environmental barriers to and facilitators of physical activity among Latinas living in San Diego County: Focus group results. Fam Community Health, 2009; 32:22–33.

33.

Dowda

, Ainsworth

, Addy

, Saunders

, Riner

. Correlates of physical activity among U.S. young adults, 18 to 30 years of age, from NHANES III. Ann Behav Med, 2003; 26:15–23.

34.

US Department of Health and Human Services. National Center for Health Statistics. Plan and operations of the Third National Health and Nutrition Examination Survey, 1988–1994. 1. Washington, DC: Centers for Disease Control and Prevention, 1996. www.cdc.gov/nchs/nhanes.htm. 2003 October 13.

35.

US Department of Health and Human Services. National Center for Health Statistics. Third National Health and Nutrition Examination Survey, 1988–1994, (NHANES III) Household Adult Data file (CD-Rom). Public use data file. Documentation #77560. Hyattsville, MD: Centers for Disease Control and Prevention, 1996. www.cdc.gov/nchs/nhanes.htm. 2003 October 13.

36.

US Department of Health and Human Services. National Center for Health Statistics. Third National Health and Nutrition Examination Survey 1988–1994, NHANES III examination data file (CD-Rom). Public use data file documentation #76200. Hyattsville, MD: Centers for Disease Control & Prevention, 1996. www.cdc.gov/nchs/nhanes.htm. 2003 October 13.

37.

Ainsworth

, Haskell

, Leon

et al. Compendium of physical activities: Classification of energy costs of human physical activities. Med Sci Sports Exerc, 1993; 25:71–80.

38.

US. Census Bureau. How the Census Bureau measures poverty. Poverty 2011. www.census.gov/hhes/www/poverty/about/overview/measure.html. 2011 December 30.

39.

Centers for Disease Control and Prevention. Body mass index. Healthy weight—It's not a diet, it's a lifestyle. 2011. www.cdc.gov/healthyweight/assessing/bmi/. 2011 December 30.

40.

Charlson

, Pompei

, Ales

, MacKenzie

. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis, 1987; 40:373–383.

41.

D'Hoore

, Bouckaert

, Tilquin

. Practical considerations on the use of the Charlson comorbidity index with administrative data bases. J Clin Epidemiol, 1996; 49:1429–1433.

42.

Schneeweiss

, Maclure

. Use of comorbidity scores for control of confounding in studies using administrative databases. Int J Epidemiol, 2000; 29:891–898.

43.

Gossett

, Jo

, Simpson

. U.S. health and nutrition: SAS® Survey Procedures and NHANES. Proceedings of the Thirty-first Annual SAS Users Group International Conference, Cary, NC. SAS Institute, Inc. 2006. www2.sas.com/proceedings/sugi31/140-31.pdf. 2009 October .

44.

US Department of Health and Human Services. Analytic and reporting guidelines: The Third National Health and Nutrition Examination Survey, NHANES III (1988–1994) Washington, DC: Centers for Disease Control and Prevention, National Center for Health Statistics www.cdc.gov/nchs/nhanes.htm. 2003 October 15.

45.

Brownson

, Eyler

, King

, Brown

, Shyu

, Sallis

. Patterns and correlates of physical activity among U.S. women 40 years and older. Am J Public Health, 2000; 90:264–270.

46.

Carlson

, Densmore

, Fulton

, Yore

, Kohl

3rd . Differences in physical activity prevalence and trends from 3 U.S. surveillance systems: NHIS, NHANES, and BRFSS. J Phys Act Health, 2009; 6,Suppl 1:S18–27.

47.

Neighbors

, Marquez

, Marcus

. Leisure-time physical activity disparities among Hispanic subgroups in the United States. Am J Public Health, 2008; 98:1460–1464.

48.

Evenson

, Sarmiento

, Tawney

, Macon

, Ammerman

. Personal, social, and environmental correlates of physical activity in North Carolina Latina immigrants. Am J Prev Med, 2003; 25,Suppl 1:77–85.

49.

Eyler

, Matson-Koffman

, Vest

et al. Environmental, policy, and cultural factors related to physical activity in a diverse sample of women: The Women's Cardiovascular Health Network Project—Summary and discussion. Women Health, 2002; 36:123–134.

50.

Eyler

, Matson-Koffman

, Vest

et al. Environmental, policy, and cultural factors related to physical activity in a diverse sample of women: The Women's Cardiovascular Health Network Project—Introduction and methodology. Women Health, 2002; 36:1–15.

51.

Eyler

, Vest

. Environmental and policy factors related to physical activity in rural white women. Women Health, 2002; 36:111–121.

52.

Abraido-Lanza

, Armbrister

, Florez

, Aguirre

. Toward a theory-driven model of acculturation in public health research. Am J Public Health, 2006; 96:1342–1346.

53.

Pekmezi

, Dunsiger

, Gaskins

, Barbera

, Marcus-Blank

, Marcus

. Feasibility and acceptability of using pedometers as an intervention tool for Latinas. J Phys Act Health, 2012[Epub ahead of print]

54.

Pekmezi

, Neighbors

, Lee

et al. A culturally adapted physical activity intervention for Latinas: A randomized controlled trial. Am J Prev Med, 2009; 37:495–500.

55.

Shephard

. Limits to the measurement of habitual physical activity by questionnaires. Br J Sports Med, 2003; 37:197–206.

56.

Prince

, Adamo

, Hamel

, Hardt

, Gorber

, Tremblay

. A comparison of direct versus self-report measures for assessing physical activity in adults: A systematic review. Int J Behav Nutr Phys Act, 2008; 5:56.