Adolescent Participation in Preventive Health Behaviors, Physical Activity, and Nutrition: Differences Across Immigrant Generations for Asians and Latinos Compared With Whites

Abstract

Objectives. We investigated preventive health behaviors (bicycle helmet, seat belt, and sunscreen use), physical activity, television viewing or video game playing, and nutrition (fruit, vegetable, milk, and soda consumption) among Asian and Latino adolescents living in the United States; assessed trends across generations (first-, second-, and third-generation immigrants or later); and compared each generation with White adolescents.

Methods. We used data from 5801 adolescents aged 12 to 17 years in the representative 2001 California Health Interview Survey.

Results. In multivariate analysis, first-generation Asians measured worse than Whites for preventive health behaviors (lower participation), physical activity (less activity), and television viewing or video game playing (more hours), but improved across generations. For these same behaviors, Latinos were similar to or worse than Whites, and generally showed no improvement across generations. First-generation Asians and Latinos had healthier diets than Whites (higher fruit and vegetable consumption, lower soda consumption). With succeeding generations, Asians’ fruit, vegetable, and soda consumption remained stable, but Latinos’ fruit and vegetable consumption decreased and their soda consumption increased, so that by the third generation Latinos’ nutrition was poorer than Whites’.

Conclusions. For the health behaviors we examined, Asian adolescents’ health behaviors either improved with each generation or remained better than that of Whites. Latino adolescents demonstrated generally worse preventive health behaviors than did Whites and, in the case of nutrition, a worsening across generations. Targeted interventions may be needed to address behavioral disparities.

In the year 2000, 1 in 5 children in the United States was an immigrant or a child of immigrants,1 yet the health and health behaviors of this large and growing population remain understudied.2 Research on immigrant adolescents suggests that rates of risk behaviors such as substance use increase across generations,3^,4 but other health-related behaviors have received less attention. Differences across racial/ethnic groups have been noted overall for prevalence of adolescent behaviors such as bicycle helmet and seat belt use, physical activity, and eating a healthy diet.5 A large percentage of the Asian and Latino populations are first- or second-generation immigrants,6 yet no studies have compared multiple generations of these groups with Whites to understand whether behavioral disparities may be emerging across generations.

Acculturation is the multidimensional and multidirectional process whereby immigrants and their descendents adopt the behaviors, beliefs, and values of the host culture while adapting those belonging to their culture of origin.7^,8 Generation status does not directly measure acculturation, but is an important and easily identifiable measure representing length of exposure to the host culture. Generation status has been associated with variations in health care access and utilization,9 education outcomes,10 and health-risk behaviors.

For Latino adolescents, substance use and sexual activity have been shown to increase with generation.11^,12 One small study that explored preventive health behaviors suggested that foreign-born Latino adolescents use seat belts more often than do US-born Latino adolescents.13 For some Latinos, the language spoken at home and the proportion of foreign-born neighbors mediate obesity-related behaviors such as dietary behavior, physical activity, and smoking, and these differences are related to the increased obesity seen in US-born Latino adolescents.14 For Asian youths, the association of behavioral trends with generation status remains largely unexplored, although one study that measured acculturation linked increased English language use to smoking.15 However, Asian adults’ use of cigarettes and alcohol increases between first and second generation,16^,17 and their diet appears to worsen.18^,19 Asian adolescents’ preventive health behaviors may similarly decline in prevalence.

The phenomenon of health and health behaviors worsening from first to later generations has been termed “the healthy immigrant effect.”20 No study has explored whether this trend holds across ethnic groups for multiple health-related behaviors such as physical activity, preventive behaviors (bicycle helmet, seat belt, and sunscreen use), television viewing and video game playing (with less considered healthier), and dietary behaviors (higher fruit, vegetable, and milk consumption; lower soda consumption). Nor have studies compared multiple racial/ethnic groups with Whites to learn whether the worsening health (with increased time in the United States) suggested by the healthy immigrant effect also signals the emergence of disparities in behaviors. Whites’ behaviors are not meant to represent an ideal, but instead to serve as a basis for comparison.

In 2002, California received more than one quarter of the nation’s immigrants—more than any other state—with Latinos making up the largest group, followed by Asians.21 Therefore, we analyzed data from a representative study of California adolescents that included a large sample of Asians and Latinos. We examined the health-related behaviors of California adolescents in 4 ways. We assessed behaviors for Asians and Latinos and compared each group as a whole first with White adolescents and secondly with each other. Next, we compared each generation with Whites. Finally, we analyzed data for within-group racial/ethnic trends across first, second, and third or higher generations for Asians and Latinos. Characterizing the health-related behaviors of immigrant populations and how these behaviors change with generation may identify subpopulations among which clinical or public health interventions may be particularly effective.

METHODS

Data Source and Participants

We used data from the adolescent portion of the 2001 California Health Interview Survey, conducted from November 2000 to September 2001. The California Health Interview Survey is a statewide telephone survey that collects information on health, health-related behaviors, and access to health care. It is representative of the state’s noninstitutionalized civilian population. Detailed methodological information appears elsewhere.22

The California Health Interview Survey used a 2-stage, random-digit-dial design. In stage 1, a random sample was taken from a list of all California telephone numbers, which were then screened to determine if they were residential households. In stage 2, 1 adult per household was randomly selected for interview. Adults who were the parent/guardian of an adolescent (aged 12–17 years) gave consent to interview 1 randomly selected youth. Adolescents provided assent. A computer-assisted telephone interview was conducted in 6 languages (Cantonese, English, Khmer, Korean, Mandarin, Spanish, and Vietnamese); 12% of adults and 9% of adolescents were interviewed in a language other than English.

Data were collected on 55 428 adults and 5801 adolescents. The household screener completion rate was 59.2%, and the overall response rate was 37.7%. This is comparable to other random-digit-dial surveys.23^,24 The adolescent interview completion rate was 63.5% among screened households. Response rates did not differ by gender or number of adolescents in the household, but parents were slightly more likely to grant permission for older (aged 15–17 years) than for younger adolescents. Poststratification and other nonresponse adjustments corrected for selective nonresponse on the basis of demographic and geographic factors.

Outcome Variables

Adolescents were asked questions on preventive health behaviors including: “How often do you wear a helmet when riding a bicycle?”; “How often do you use a seat belt when riding or driving in a car?”; and “How often do you use strong sunscreen—a sun-screen with sun-protection factor of 15 or greater—when you go outside on a very sunny day for more than 1 hour?” The median was used to dichotomize these variables. Bicycle helmet use was calculated only for respondents who rode bicycles, and was coded as 1 = always or usually wears a helmet versus 0 = sometimes or never. Seat belt use was coded as 1 = always uses a seat belt and 0 = usually, sometimes, or never. Sunscreen use had 3 responses, and was coded as 1 = always and 0 = sometimes or never.

Adolescents were asked about physical activity: “In the past 7 days, did you do any physical activity for at least 20 minutes that made you sweat and breathe hard, such as basketball, soccer, running, swimming laps, fast bicycling, fast dancing, or similar aerobic activities?” They were also asked: “Again, thinking about the past 7 days, did you do any physical activity for at least half an hour that did not make you sweat or breathe hard? This could be things like walking for at least 30 minutes, slow bicycling or skating, or doing chores like cleaning the house or mopping floors.” Responses for both questions ranged from 0 to 7 days. The percentage meeting 2004 Centers for Disease Control and Prevention adolescent recommendations for minimal activity was calculated with the following standards: vigorous activity for at least 20 minutes or moderate activity for at least 30 minutes over at least 3 of the past 7 days. This definition appeared on the Centers for Disease Control and Prevention’s public education Web site before adoption of new recommendations in 2005,25 and was recommended in the 2000 federal report to the President on physical activity.26

Adolescents were asked, “Thinking about your free time on Monday through Friday, on a typical day, about how many hours do you usually watch TV or play video games (such as PlayStation)?” The question was repeated for Saturday and Sunday. The average daily number of hours spent on these activities was calculated as a weighted average of weekday and weekend responses.

Mean daily dietary intake was calculated from: “How many servings of fruit, such as an apple or banana, did you have yesterday?”; “How many servings of vegetables, such as corn, green beans, lettuce, or other vegetables, did you have yesterday?”; “How many glasses of milk did you drink yesterday?”; and “How many glasses or cans of soda, such as Coke or 7-Up, did you have yesterday?”

Independent Variables

Adolescents were defined as first generation if they were born outside the United States, second generation if US-born with 1 or more foreign-born parents, and third generation or later if the adolescent and both parents were born in the United States. Adolescents were asked to describe themselves as Latino/Hispanic or not, and then as 1 or more of the following: American Indian or Alaska Native; Asian; Black or African American; Native Hawaiian or other Pacific Islander; White; or other. Respondents choosing multiple groups were asked which group they most identified with. Respondents who again chose multiple groups were categorized as “other or multiple races.” Age and gender were self-reported. The parent or guardian reported his or her own level of education (less than high school, high school, some college, college graduate, and greater), household income, and number of household members (from which percentage of federal poverty level was calculated), and residence zip code (population density by zip code was used to determine rural vs urban residence27).

Data Analysis

Descriptive statistics were computed by race/ethnicity and generation within Asian and Latino groups. These included means for nutrition and television or video game outcomes (which were continuous) and percentages for other outcomes (which were dichotomous).

A first set of multivariate regressions predicted outcomes from a model of racial/ethnic and generation indicators (within Asian and Latino groups). Whites were the reference group. The models made 3 sets of comparisons: (1) each racial/ethnic group (i.e., Asians, Latinos) versus Whites, and versus each other (pooling across generation for each); (2) each Asian and Latino generation versus Whites overall; and (3) a test of within-group racial/ethnic generation trend among Asians and Latinos. These regressions were linear for continuous outcomes and logistic for dichotomous outcomes.

A second set of multivariate models predicted the same outcomes when adjusted for the following covariates: adolescent age and gender, parent education level, rural versus urban residence, and percentage of federal poverty income level. These models are referred to as the adjusted models. Using recycled predictions,28 we present covariate-adjusted means from linear regressions and covariate-adjusted percentages from logistic regressions by race/ethnicity and generation.

Final analyses were weighted, and SAS version 9.1 (SAS Institute Inc, Cary, NC) survey procedures that account for the California Health Interview Survey’s complex sample design were used. No variable was missing for more than 2% of cases. Because missing responses were rare, we used mean imputation for continuous variables, median imputation for ordered categories, and modal imputation for unordered categories.

Data for all racial/ethnic groups are presented in Table 1 ▶ and indicated in the “Total” lines in Tables 2 ▶ and 3 ▶, but regression results are presented only for Asians and Latinos.

TABLE 1—

Percentage Distribution by Selected Characteristics: California Adolescents, California Health Interview Survey, 2001

	Number (N=5801)	%a
Race/Ethnicity
African American	308	7
AIAN	115	1
White	3263	44
Other	224	4
Asian	376	8
First generation	122	36
Second generation	202	52
Third generation	52	12
Latino	1515	36
First generation	401	28
Second generation	768	56
Third generation	346	16
Sociodemographics
Age
12–13	1963	34
14–15	1937	33
16–17	1901	33
Gender
Boy	2905	51
Girl	2896	49
Parent’s educationb
Less than high school	1024	25
High-school graduate	1400	23
Some college	1661	25
College degree or higher	1716	27
% of federal poverty level
0–99	878	21
100–199	1119	21
200–299	963	15
≥ 300	2841	43
Residence
Urban	4252	86
Rural	1549	14

Notes. AIAN = American Indian or Alaska Native.

^a Values are expressed as percentages. Numbers are unweighted, percentages are weighted.

^b Mother, father, or adult guardian.

TABLE 2—

Participation of Asians and Latinos in Preventive Health Behaviors and Physical Activity, Difference From Whites, and Generation Trends: California Adolescents, California Health Interview Survey, 2001

	Used Bicycle Helmeta		Always Used Seat Belt		Used Sunscreenb		Physical Activityc
	Unadjusted %	Adjusted %	Unadjusted %	Adjusted %	Unadjusted %	Adjusted %	Unadjusted %	Adjusted %
Total	50		80		57		73
White	65	60	82	80	69	66	77	76
Asiand,e,f	56*	53	71***	71**	57**	56**	66**	67**
First generationg	45**	45*	58***	60***	44***	48**	57***	58***
Second generationg	58	53	75	74	64	60	70	70
Third generationg	75	69	95	94	70	65	81	80
Latinod,f,h	29***	36***	79	83	48***	57***	70***	72
First generationg	29***	41***	82	87*	47***	60	68**	70
Second generationg	27***	34***	78	82	48***	58**	69**	72
Third generationg	34***	35***	78	79	49***	52***	74	75

Notes. Values are predicted percentages based on logistic regression models. Adjustment was made for gender, age, parent’s education, percentage of federal poverty level, and rural versus urban residence. Levels of missing responses were less than 2% for all outcomes.

^aAlways/sometimes. N = 4609.

^bAlways/sometimes.

^c Moderate activity for 30 minutes or vigorous activity for 20 minutes at least 3 of the past 7 days.

^d Test of race/ethnicity versus White.

^e Overall F test for generational trend among Asians was significant at P < .05 for physical activity (unadjusted model) and at P < .01 for always using a seat belt (both adjusted and unadjusted models) and using sunscreen (unadjusted model).

^f The test of differences between Asians and Latinos overall was significant at P < .05 for using sunscreen (unadjusted model), at P < .01 for always using a seat belt (unadjusted model), and at P < .001 for using a bicycle helmet (both adjusted and unadjusted models) and always using a seatbelt (adjusted model).

^g Test of generation versus Whites.

^h Overall F test for generational trend among Latinos was not significant for any model.

* P < .05; **P < .01; ***P < .001.

TABLE 3—

Participation of Asians and Latinos in Television Viewing or Video Game Playing and Consumption of Foods, Difference From Whites, and Generation Trends: California Adolescents, California Health Interview Survey, 2001

	Hours of Television/Video Gamesa		Servings Fruitb		Servings Vegetableb		Glasses Milkb		Glasses/Cans Sodab
	Unadjusted	Adjusted	Unadjusted	Adjusted	Unadjusted	Adjusted	Unadjusted	Adjusted	Unadjusted	Adjusted
Total	2.7		1.5		1.3		1.5		1.2
White	2.5	2.6	1.5	1.5	1.3	1.3	1.6	1.5	1.2	1.2
Asianc,d,e	2.8**	2.9***	1.6	1.5	1.6***	1.5***	1.3***	1.3**	0.8***	0.8***
First generationf	2.8	2.8*	1.7	1.7*	1.6**	1.6***	1.0***	1.1***	0.8***	0.8***
Second generationf	2.9**	2.9**	1.5	1.4	1.5**	1.5**	1.3**	1.3**	0.8***	0.9***
Third generationf	2.5	2.6	1.7	1.6	1.7**	1.7*	2.0**	2.1**	0.8*	0.9*
Latinoc,e,g	2.8***	2.7	1.6	1.5	1.2**	1.2	1.5*	1.5	1.2	1.1*
First generationf	2.8***	2.7	1.8***	1.8***	1.4	1.4*	1.4***	1.5	1.1	0.9***
Second generationf	2.7***	2.6	1.6	1.6	1.2**	1.2	1.5	1.6	1.2	1.1**
Third generationf	2.9***	2.8*	1.1***	1.2***	1.0***	1.0***	1.4*	1.4*	1.5***	1.5*

Notes. Values are predicted means based on linear regression models. Adjustment was made for gender, age, parent’s education, percentage of federal poverty level, and rural versus urban residence. Levels of missing responses were less than 2% for all outcomes.

^a Average daily time spent watching television or playing video games in hours.

^b Number of servings or glasses reported the day preceding the interview.

^c Test of race/ethnicity versus Whites.

^d Omnibus F test for generational trend among Asians was significant at P < .001 for number of glasses of milk consumed (both adjusted and unadjusted models).

^e The test of differences between Asians and Latinos overall was significant P < .05 for hours of television watched or video games played (adjusted model), at P < .01 for number of glasses of milk consumed (both adjusted and unadjusted models), and at P < .001 for number of servings of vegetables consumed (both adjusted and unadjusted models) and number of glasses or cans of soda consumed (both adjusted and unadjusted models).

^f Test of generation versus Whites.

^g Omnibus F test for generational trend among Lations was significant at P < .05 for number of glasses of milk consumed (unadjusted model), at P < .01 for number of glasses of milk consumed (adjusted model), and at P < .001 for servings of fruits consumed (both adjusted and unadjusted models), servings of vegetables consumed (both adjusted and unadjusted models), and glasses or cans of soda consumed (both adjusted and unadjusted models).

* P < .05; **P < .01; ***P < .001.

RESULTS

General Characteristics

The sample of 5801 participants was weighted to reflect the composition of California’s adolescents aged 12 to 17 years (Table 1 ▶). The racial/ethnic distribution included White (44%), Latino (36%), and Asian (8%) adolescents. Most Asians and Latinos were first (36% and 28%, respectively) or second (52% and 56%, respectively) generation.

Preventive Behaviors

Overall, Asians were less likely than Whites to use bicycle helmets (P = .05), seat belts (P = .002), and sunscreen (P = .002) (Table 2 ▶). For all 3 preventive behaviors, participation for first-generation Asians was lower than for Whites (P = .004 for bicycle helmet, P < .001 for seat belt and sunscreen), but were not significantly different from Whites in later generations. These patterns remain true for all 3 behaviors after adjustment for socioeconomic characteristics despite the loss of the overall generation trend for sunscreen use.

Latinos were less likely than were Whites to use bicycle helmets (P < .001) and sun-screen (P < .001), but as likely to use seat belts. Latino bicycle helmet use for all 3 generations remained lower than use among Whites even after adjustment (P < .001 for each generation). Adjusted analyses found that first-generation Latinos (but not later generations) were more frequently users of seat belts than were Whites, and similar to Whites in sunscreen use. Latinos showed no intragroup generational trends for any of the outcomes.

In comparisons between Asians and Latinos, Asians were more frequent bicycle helmet (P<.001) and sunscreen users (P=.01) before adjustment. Overall, Latinos were more frequent seat belt users. However, in generational comparisons between the groups, although first-generation Latinos were more frequent users than were first-generation Asians (P<.001, data not shown), third-generation Latinos and Asians were similar in usage frequency.

Physical Activity and Television Viewing or Video Game Playing

Overall, fewer Asians than Whites met physical activity guidelines, but there was an unadjusted trend toward increased participation (P=.04) across generations. Although first-generation Asians were significantly less active than Whites (P<.001), second- and third-generation Asians were not significantly different from Whites. The generation trend toward increased physical activity was not significant in the adjusted models. Fewer Latinos met the activity recommendations than Whites in general (P<.001) and for first and second generations in particular (P=.003 and P=.004, respectively), but they were similar at third generation. After adjustment, the difference between Latinos and Whites was not significant overall or at each of the generations. There were no intragroup generational trends.

Asians reported more hours of television viewing and video game playing than Whites (P = .002) (Table 3 ▶). In adjusted models, first-and second-generation Asians reported more hours than Whites (P = .05 and P = .001, respectively), but at third generation they were similar. Latinos reported more hours overall (P < .001) and at each generation (P < .001 for all) than Whites before adjustment, but after adjustment only third-generation Latinos remained higher than Whites (P = .02). There were no intragroup generational trends for either Asians or Latinos. After adjustment, Asians’ television viewing or video game playing was significantly higher than Latinos’ (P = .04).

Nutrition

Compared with Whites, Asians consumed more servings of vegetables daily (P < .001), fewer servings of soda (P < .001), and similar amounts of fruit (Table 3 ▶). There were no intragroup generational trends for these outcomes. Adjusted fruit consumption in first-generation Asians was slightly higher than for Whites, but was not significantly different in second or third generations. Vegetable consumption was higher at each generation and soda consumption was lower in both unadjusted and adjusted models. Overall, Asian milk consumption was lower than Whites’ (P < .001), but the intragroup generational trend suggested improvement (P < .001). Asians had lower milk consumption than Whites in first and second generations (P < .01 for both) but surpassed them by third generation (P = .002). This trend represents a doubling of milk consumption from first to third generation.

Overall, Latino youth were similar to Whites for fruit and soda consumption, and reported less vegetable and milk consumption. Fruit, vegetable, and soda consumption all showed intragroup generational trends (P < .001) in the direction of worsening nutrition. First-generation Latinos consumed more fruit (P < .001) and similar amounts of vegetables compared with Whites, but by the third generation consumed fewer fruits and vegetables (P < .001 for both). First- and second-generation Latino youth were similar to Whites for soda consumption, but third-generation Latinos consumed about 25% more soda than Whites (P = .01). After adjustment, first-generation Latinos were no longer significantly different from Whites in milk consumption; first and second generations consumed less soda than Whites of similar socioeconomic status, but third generation consumed more. The generation trends remained significant.

Comparisons between Asians and Latinos showed no differences in fruit consumption. Asians consumed vegetables more (P < .001), and Latinos consumed soda more (P < .001). Overall, Latinos consumed more milk than Asians, but in generational comparisons between the groups (data not shown), third-generation Asians consumed more milk (P < .001).

As a sensitivity test, we examined whether our overall comparisons by race/ethnicity would have differed if the reference group had been restricted to the 88% of Whites classified as third generation or higher. Because we found no evidence of generation differences among Whites at P < .05 except for 3 dietary behaviors (fruit, vegetable, and soda consumption), we restricted our sensitivity tests to these 3 outcomes; the significance of reported results did not change at the 5% threshold for any of them.

DISCUSSION

Our study illustrates distinctive patterns of health-related behaviors across immigrant generations for Asian and Latino adolescents in California, even after control for socioeconomic status. Generally, Asians showed maintenance or improvement of the health behaviors examined in this article across generations, whereas Latinos showed either poorer preventive health behaviors than Whites or a worsening of behaviors across generations. Similarly, we found differences between the 2 populations in the prevalence of disparities in these health behaviors when compared to Whites. When present, disparities for Asians appeared in the first generation but not in later generations, whereas for Latinos disparities either appeared in later generations or persisted across generations. These results suggest that the association of behavior with the length of exposure to the United States mainstream culture as measured by generation status may differ by immigrant population.

Latinos showed no trends across generations toward improving disparities compared with Whites in bicycle helmet and sunscreen use, and they showed trends in the direction of worsening nutrition for fruit, vegetable, and soda consumption. Given the generational trend from better to poorer nutrition than Whites, Latino youth exhibit an emerging disparity in these health behaviors across generations. These results are consistent with other studies that suggested that acculturation has a negative effect on measured health-risk health behaviors in this population.11^–13 The exception in this study was in meeting Centers for Disease Control and Prevention physical activity recommendations, for which, similar to findings in previous studies,14 minimal variation in levels of physical activity were noted across generations.

The generational trend noted for Latino youth in nutrition is particularly concerning because it may contribute to the high rates of obesity in this community.29 Interventions targeting Latino youth who are new immigrants may help to reinforce healthful nutritional habits.

Unlike previous research suggesting that Asian adolescents’ health-related behaviors might worsen across generations,15^,17^–19 our study found that their behaviors improved. In our study, elimination of the disparities seen in first-generation Asian adolescents was noted for bicycle helmet, seat belt, and sun-screen use; percentage meeting Centers for Disease Control and Prevention physical activity recommendations; and hours of television viewing or video game playing. Asians also showed either improvement or maintenance of better nutrition compared with Whites. The improvement in health-promoting behaviors across generations may be because this is the first large and representative sample to study generational trends for Asian adolescents, or the particular behaviors that were assessed. It is possible that other behaviors such as substance use or sexual activity might show different patterns. Our findings may be limited by the small sample size for third-generation Asians, decreasing our ability to note differences between this group and Whites. Because the generational trends included all Asians, we can be assured by the presence of trends for most preventive behaviors and absence of trends for most dietary behaviors that the patterns noted here would likely hold if we did have more statistical power.

Little research has examined why one immigrant population might meet or surpass the majority population in health-related behaviors (e.g., Asians’ preventive behaviors), while another loses its relative behavioral health advantage (e.g., Latinos’ dietary behaviors). Evidence on other topics is informative. Research that utilized the theory of segmental assimilation identified potential factors leading to wide variation across racial/ethnic groups in educational aspiration and attainment of second-generation children.30 These factors include characteristics of the immigrant group such as average education and job skill levels, the proportion who were documented immigrants or refugees, and the proportion living in 2-parent households. Barriers to achievement included perceived racism and the more rapid acculturation of teens than their parents. Community characteristics thought to improve outcomes in this study included high levels of social capital. Research is needed to determine whether factors that influence educational attainment also influence preventive health behaviors and explain differences among racial/ethnic groups. Further insight comes from an operant theory of acculturation; this theory predicts that as immigrants acculturate, health behaviors from their culture of origin that are not reinforced by the dominant culture will wane, whereas those that are reinforced will increase.8 In our study, findings for Asians are consistent with this model in terms of preventive health behaviors, and for Latinos in terms of fruit and vegetable consumption.

Because we have analyzed cross-sectional data, we must be cautious in interpreting associations as causal relationships. The data are from California, which may limit generalizability to areas with a different mix of ethnic subgroups. Although we controlled for some socioeconomic factors, cultural variation within subgroups could contribute to generational differences in behaviors. For example, recent Asian immigrants are more likely to come from South, Central, or Southeast Asia and more likely to have lower socioeconomic status than in previous decades.31^,32 Sample size restricted our ability to perform ethnic subgroup analyses.

Despite the variation by race/ethnicity and generation, our study also points out that all groups’ preventive health behaviors might benefit from clinical and public health interventions. All groups consumed a substantial amount of soda and fewer servings of fruit, vegetables, and milk than recommended for a healthy diet.33^,34 There is significant variation in bicycle helmet use among racial/ethnic groups, but even among Whites—the highest users—more than a third of adolescents rarely or never use helmets, placing them at risk of head injury.5 Disparities in bicycle helmet use have not been explored in the research literature, and although general barriers to helmet use have been identified,35 further research is needed to understand how best to reduce the gap.

All youths spent more time watching television and playing video games per day than the 2-hour maximum recommended by the American Academy of Pediatrics.36 There is not, however, a clear consensus on what are reasonable cutoffs for the amount of time spent with these activities, and more research is needed to determine the nature of any associations with risks for obesity and other health problems. Sunscreen use was infrequent. Risk for skin cancers is higher for lighter-skinned populations, but the American Academy of Dermatology37 and the American Academy of Pediatrics38 recommend daily use of sunscreen with a sun-protection factor of 15 or greater for all racial/ethnic groups.

Our findings suggest that, across generations, Latinos show persistent disparities for preventive behaviors and increasing disparities for dietary behaviors, but that for Asians many disparities decrease. Further investigation is required to understand why these differing trends occur, and how interventions may best reinforce positive health behaviors that immigrants bring from their home countries.