Physical Activity Correlates for Native Hawaiian and Pacific Islanders in the United States

Abstract

Native Hawaiians and Pacific Islanders (NHPI) are an understudied population that demonstrates high obesity rates and low physical activity levels. This study’s aim was to examine possible correlates of physical activity in NHPI adults. Height and weight were recorded in N=100 NHPI (46.9±5.4 years; 56% males) following completion of an anonymous questionnaire addressing health behaviors (physical activity, smoking, diet), psychosocial variables (social support, barriers, stage of change), neighborhood environment attributes, and knowledge of physical activity recommendations. This study sample demonstrated low physical activity (20% met recommendations) and fruit and vegetable (F&V) consumption (1% met recommendations), and a high prevalence of overweight or obesity (94%). After adjusting for gender and education, F&V intake was the only significant correlate of physical activity (p<0.001). Common correlates of physical activity did not generalize to NHPI. Further investigations of culturally-specific correlates are needed so that physical activity interventions can be culturally tailored for NHPI.

Regular physical activity is recognized internationally as a behavior important for the prevention and treatment of several major chronic diseases and conditions. Adults can reap substantial health benefits by performing at least 150 minutes per week of moderate-intensity, or 75 minutes per week of vigorous-intensity aerobic physical activity, or an equivalent combination of moderate- and vigorous-intensity activity.¹

Native Hawaiian and Pacific Islanders (NHPI) are at much greater risk than other Americans for inactivity-related diseases, including obesity, cardiovascular disease, non-insulin-dependent diabetes mellitus, and some cancers.^2–4 Thus, it is of concern that NHPI are typically found to be less active than their Caucasian counterparts, and a large proportion of NHPI do not meet current physical activity recommendations for health benefits.^3,5–10 Despite the 1997 update on race and ethnicity classification that separated Asian Americans and NHPI,¹¹ physical activity prevalence data specific to NHPI populations remain scarce. In an NHPI adult sample in San Diego, 80% were not meeting current physical activity recommendations to promote or maintain health¹² compared to 16% of the general U.S. population,¹³ based on the same questionnaire. Another study reported that nearly half of NHPI in California engaged in less than 10 minutes per week of walking for leisure or transportation.²

The underlying reasons behind typically low activity levels among NHPI are unknown. Many variables can impact an individual’s physical activity behavior, including demographic factors such as age, gender, race, education, and household income.^3,14,15 Social and environmental factors can act as either facilitators or barriers to this behavior. For example, social support is consistently correlated with physical activity.^14,16–18 Neighborhood environment variables such as having recreational facilities within close proximity and highly walkable neighborhoods are also consistently related to activity levels.^19,20

Several health-related behaviors are associated with physical activity.^18,21,22 A healthy diet, defined as adequate fruit and vegetable (F&V) consumption and avoidance of high-fat foods, is positively associated with physical activity levels.^14,21,22 The association between smoking and physical activity is less consistent, with some studies showing a lack of association^14,18,22 and others showing an inverse relationship with physical activity.^21–24

These health behavior, psychosocial, and environmental variables should be explored as physical activity correlates specifically among NHPI. Improved understanding of physical activity correlates among NHPI can provide an evidence base for culturally-tailored, community-based interventions to be created and sustained. The primary aim of the present study was to examine correlates of physical activity for NHPI adults.

Methods

This cross-sectional study was approved by the San Diego State University Research Foundation Institutional Review Board. A convenience sample of NHPI adults between ages 40–59 years were recruited by a community leader during the San Diego Pacific Islander Festival and local church and community functions in 2007.

The anonymous questionnaire utilized questions from existing surveys that had established validity and reliability and consisted of 59–62 questions, depending on sex and physical activity level. Table 1 illustrates this study’s variables of interest, the instrument from which each set of questions was taken, the number of items for each variable, sample items and Cronbach’s alpha. After interviewer-administration of the physical activity questions, the remainder of the questionnaire was self-administered. A digital scale (Conair WW33) was used to measure weight to the nearest 0.1 kg and a stadiometer (Ohaus ES 200L) was used to measure height to the nearest 0.1 cm. Body mass index (BMI) was calculated as weight (kg) divided by height (m²) and standard BMI categories were used to categorize participants as normal weight (<25.0 kg/m²), overweight (25.0–29.9 kg/m²) or obese (≥30.0 kg/m²).

Table 1.

Composition of questionnaire to determine health behaviors and correlates of Native Hawaiians and Pacific Islanders

Variable	Original Instrument	# of items	Description, sample items	Cronbach’s alpha
Education	Behavioral Risk Factor Surveillance System 200629	1	Highest education level completed  1 = Less than 7th grade,  7 = Completed graduate degree	–
Annual household income	Behavioral Risk Factor Surveillance System 200629	1	Annual household income:  1 = <$10,000; 11 = >$100,000	–
Language spoken at home	U.S. Census 200040	1	English spoken at home: 1 = Yes; 2 = No	–
Physical activity	New Zealand Physical Activity Questionnaire – short form9	5–8	Duration and frequency of walking, moderate- and vigorous-intensity activity during the last 7 days; stage of change	–
Fruits & vegetable consumption	Behavioral Risk Factor Surveillance System 200530	6	Usual consumption of fruits and vegetables (fruit, fruit juice, green salad, potatoes, carrots and vegetables)	0.59
Tobacco use	Behavioral Risk Factor Surveillance System 200629	2	Ever smoker: 1 = Yes, 2 = No Frequency: 1 = Every day, 2 = Some days	–
Knowledge of physical activity recommendations	Centers for Disease Control and Prevention/American College of Sports Medicine33	7	What is the minimum number of days per week a person must be physically active in order to receive any health benefit?: 0 = incorrect or “I don’t know”1 = correct	–
Neighborhood environment	Physical Activity Neighborhood Environment Scale32	15	The crime rate in my neighborhood makes it unsafe to go on walks at night: 1 = Strongly disagree; 4 = Strongly agree	0.77
Social support for exercise	Neighborhood Quality of Life Study31	6	During the past 3 months my family or friends offered to do physical activity with me: 0 = Never; 4 = Very often	0.80 (family) 0.75 (friends)
Barriers to physical activity	Neighborhood Quality of Life Study31	15	How often does ‘lack of time’ prevent you from getting regular physical activity?: 0 = Never; 4 = Very often	0.74

The short form of the New Zealand Physical Activity Questionnaire (NZPAQ-SF)²⁵ was used to assess the duration and frequency of brisk walking, moderate- and vigorous-intensity activities performed in the last seven days. The NZPAQ-SF, an adaptation of the International Physical Activity Questionnaire (IPAQ), was validated against heart rate monitoring in a multiethnic population, including Pacific Islanders, and demonstrated acceptable validity (r=0.25, p<0.001).⁹ The products of self-reported frequency (days/week) and average daily duration (min/day) of walking, moderate- and vigorous-intensity activities were summed to calculate total physical activity (min/week). Corresponding metabolic equivalent (MET) values for walking, moderate- and vigorous-intensity activity (3.3, 4.0, and 8.0 METs, respectively) were multiplied by duration of each activity (min/week). The sum of these values produced total MET-min/week of physical activity based on scoring criteria established by the IPAQ Committee.²⁶ Stage of behavior change was assessed by asking participants to describe their physical activity levels over the last six months. Responses were categorized into one of five readiness-to-change categories: precontemplation, contemplation, preparation, action and maintenance.²⁷

Participants were categorized either as current smokers or non-smokers,²⁸ and responses to dietary questions were converted to average daily servings of F&V, as used in the Behavioral Risk Factor Surveillance Surveys.^29,30 The social support scales addressed support from family and friends.³¹ For some analyses, responses to social support and barriers to regular physical activity were aggregated into two categories: rarely (‘never’ and ‘rarely’) and sometimes or more often (‘sometimes’, ‘often’, and ‘very often’). The Physical Activity Neighborhood Environment Scale (PANES)³² addressed neighborhood characteristics believed to support physical activity, and for some analyses, responses were aggregated into two categories: agree (‘somewhat agree’ and ‘strongly agree’) and disagree (‘somewhat disagree’ and ‘strongly disagree’). Knowledge of physical activity recommendations (KPAR) were analyzed as mean KPAR scores the percentage of correct responses for each question were analyzed.³³

Statistical analyses

A target sample size of N=100 was set to detect small effect sizes (Cohen’s d = 0.20) at an alpha of 0.05 and power of 0.80. All statistical analyses were conducted using SPSS v15.0, and ‘Don’t know/Not sure’ and ‘Refused’ responses were set to “missing”. All continuous variables were checked for normality assumptions (including skew, kurtosis, and outliers). Percent endorsement for each item is presented. Correlates of physical activity were examined using Pearson correlations between physical activity and other continuous variables. Point bi-serial correlations are reported for physical activity and dichotomous variables. If significant correlations were found (p<0.05), they were included as independent variables into a hierarchical linear regression, with total MET-min/week as the dependent variable to determine the extent, if any, to which demographic, psychosocial, neighborhood environment or knowledge variables explained physical activity. All variables included in the regression were examined for multi-collinearity and outliers. Demographic questions were entered stepwise into the first model (education and gender). All other significant variables from bivariate analyses (social support from family, social support from friends, barriers to physical activity, neighborhood environment, and F&V consumption) were entered into the second model.

Results

The sample of N=100 (56M, 44F) NHPI adults had a mean age of 46.9±5.4 years and represented the main U.S. NHPI subgroups, as reported elsewhere.¹² Descriptives of all variables are reported in Table 2. The majority of participants had ‘at least a high school education’ (85.0%) and reported an annual household income of >$50,000 (70.0%). The total sample had a mean BMI of 33.8±7.5 kg/m². Only 6.0% of the sample was normal weight, while 94.0% was overweight or obese. Females (47.7%) were more likely to be overweight than males (39.3%), while males were more likely to be obese (43.2% vs. 57.1%, respectively).

Table 2.

Study variable descriptives and correlations with physical activity

Demographics	Total (N=100)	Males (n=56)	Females (n=44)	Correlation with Physical Activity
Age, y, Mean ± SD	46.9 ± 5.4	47.2 ± 5.0	46.4 ± 5.9	−.11
Gender	–	–	–	.35***
Education	High school	High school	Some college	.28**
Mean household income	$50–59K	$50–59K	$50–59K	−.03
Mean body mass index, kg/m2	33.8 ± 7.5	34.0 ± 6.4	33.7 ± 8.8	−.01
Physical Activity Levels		Mean ± SD
Total	372.6 ± 517.3	213.4 ± 334.2***	575.1 ± 631.1	–
Walking	37.26 ± 56.07	18.1 ± 35.5***	61.6 ± 67.4	–
Moderate-intensity	32 ± 58.67	18.6 ± 29.9*	49.1 ± 79.1	–
Vigorous-intensity	15.20 ± 40.18	9.9 ± 28.7	21.9 ± 50.8	–
Health Behaviors
Current tobacco use, %	49.0	51.8	45.5	.04
Daily fruit and vegetable servings, Mean ± SD	1.0 ± 1.4	0.71 ± 1.0	1.28 ± 1.7	.55***
Psychosocial, Environmental and Knowledge Variables		Mean ± SD
Stage of behavior change	2.20 ± 1.26	1.93 ± 1.23*	2.55 ± 1.23	.69***
Barriers	1.59 ± 0.46	1.66 ± 0.41	1.49 ± 0.51	−.30**
Social support – total	1.05 ± 0.70	0.84 ± 0.67**	1.31 ± 0.64	.33**
Social support – family	1.06 ± 0.83	0.85 ± 0.86**	1.32 ± 0.72	.26**
Social support – friends	1.04 ± 0.82	0.83 ± 0.77**	1.30 ± 0.81	.30**
Neighborhood environment	2.43 ± 0.46	2.32 ± 0.44	2.57 ± 0.45	.27**
Mean knowledge of physical activity score	11.9 ± 4.0	12.4 ± 4.0	11.3 ± 4.1	−.03

^*p<0.05;

^**p<0.01;

^***p<0.001

The mean self-reported physical activity exceeded the recommended 150 min/week, but a surprising finding was that physical activity reported by NHPI women were more than double activity levels reported by men (p<0.001). Half the sample reported current tobacco use (49.0%). Mean daily servings of F&V were well below the recommended minimum of five servings per day (1.0±1.4). Females reported higher daily F&V consumption than males, which approached significance (p=0.07).

Responses to the stage of behavior change question revealed that the majority of participants were pre-contemplators (36.0%), meaning they were not currently active on a regular basis and did not intend to change in the next six months. The remainder of the sample was distributed as follows: 23.0% contemplators, 22.0% preparation, 10.0% action, and 6.0% maintenance. In this sample of pre-contemplators, ‘lack of self-discipline’ was the primary barrier (80.0%), followed by ‘lack of interest’ (78.0%), ‘lack of company’ (76.0%), ‘lack of energy’ (73.0%) and ‘lack of time’ (70.0%). Participants were more likely to receive social support from friends and family ‘at least sometimes’ in the form of ‘encouragement to do physical activity’ (43.0–54.0%) compared to ‘offering to do physical activity’ (37.0–39.0%) or ‘doing physical activity’ (20.0–22.0%) with them.

The majority of participants reported having multiple shops, markets, and stores (82.0%), as well as a transit stop (70.0%) within easy walking distance from their home. The presence of several 4-way intersections (73.0%), sidewalks on most streets (75.0%) and having interesting things to look at (69.0%) was also common. However, many participants stated that traffic issues made it difficult or unpleasant to walk in their neighborhoods (66.0%), and that crime rates deterred them from walking during the day (39.0%) and night (55.0%). Less than half the participants reported having several free or low-cost recreation facilities (40.0%), well-maintained sidewalks (45.0%) or bike paths (37.0%), or seeing their neighbors being physically active (44.0%) in the neighborhood.

Overall, participants’ mean KPAR score was 11.9±4.0 out of 18. The majority of this NHPI sample knew that ‘everyone should get 30 minutes of moderate-intensity physical activity most days of the week (89.0%), and that ‘10 minutes of physical activity three times per day provides the same health benefits as a single session of 30 minutes’ (69.0%). Three-quarters (74.0%) of the sample incorrectly labeled the statement ‘vigorous levels of physical activity are necessary to provide a health benefit’ was true. Almost half (47.0%) the participants correctly identified at least 10 out of 12 activities that provide a health benefit.

Correlates of physical activity

As shown in Table 2, gender, education, F&V consumption, stage of behavior change, barriers, social support and neighborhood environment were all significantly correlated with physical activity (p<0.01). Although the stage of change measure was strongly correlated (0.69) with the physical activity interview score, this variable was considered an alternative physical activity measure, so deleted from the model.

Both models from multiple hierarchical linear regression analyses significantly explained physical activity (Model 1: Adj. R²=0.15, F(2, 91)=8.98, p<0.001; Model 2: Adj. R²=0.33, F(7, 91)=7.32, p<0.001). After controlling for gender and education in Model 1, the inclusion of Model 2 variables produced a significant R square change (R² change=0.21, F(5, 84)=5.70, p<0.001). Table 3 includes regression coefficients, t-test and part correlations for all variables in Model 2. When the effects of all other variables were taken into account, only F&V intake significantly explained physical activity levels (p<0.001).

Table 3.

Hierarchical regression results for individual variables entered in Model 2 explaining physical activity

Variable	B	B	t-test	p-value	Part r
Education	73.16	0.13	1.34	0.18	0.12
Gender	222.69	0.21	2.25	0.08	0.19
Social Support – family	4.90	0.01	0.08	0.94	0.01
Social Support – friends	37.11	0.06	0.53	0.60	0.05
Barriers	−131.75	−0.12	−1.25	0.21	−0.11
Neighborhood Environment	−48.63	−0.04	−0.42	0.68	−0.04
Fruit and Vegetable Consumption	161.72	0.43	4.43	0.000	0.38

Discussion

The primary finding of this study was that, after adjusting for gender and education, fruit and vegetable intake was the only significant correlate of physical activity for NHPI adults, with a substantial partial r of 0.38. The strong correlation between physical activity and F&V intake suggests a multiple-behavioral intervention could be particularly appropriate for NHPI adults.

The psychosocial and environmental variables were all significantly correlated with physical activity, as expected, in bivariate analyses. Bivariate physical activity correlates reported for this sample of NHPI adults were generally consistent with the literature.^14,18,34 Positive correlations were observed between physical activity and female gender, education, F&V intake, stage of behavior change, social support and neighborhood environment, while barriers were negatively correlated. Interestingly, correlations between physical activity and income or BMI were not statistically significant, but the lack of correlation with the physical activity knowledge variable was consistent with other studies.^14,18,34,35 Multi-collinearity did not appear to account for the lack of significant findings in regression analyses, but limited power due to sample size likely played a role. Because of these surprising regression findings, replications are needed with larger samples.

Although mean total physical activity levels exceeded current guidelines, when the data were scored to estimate the proportion of the sample meeting guidelines, 80.0% of the sample was classified as insufficiently active, as reported in another paper based on this sample.¹² Although males are typically more active than females,^18,36,37 the reverse was found in this sample. Self-reported walking, moderate- and vigorous-intensity activity levels for NHPI females were more than double those reported by their male counterparts. This study’s sample reported very low F&V consumption (less than one serving per day), similar to previous findings on NHPI samples.^38,39 Low prevalence of physical activity and F&V intake provide an additional rationale for interventions targeting multiple health behaviors for NHPI.

This study is the first to explore a comprehensive set of correlates of physical activity in U.S. NHPI, an under-studied population that has demonstrated high prevalence of obesity and increased risk for chronic diseases.^2,12,38 This study is not without limitations. Participants were a small convenience sample recruited by NHPI representatives through flyers and word-of-mouth, and therefore consisted of volunteers interested in taking part in a health-related research study. Although the self-report physical activity instrument used in this study had established validity in NHPI overseas, it would be useful to document reliability and validity in diverse NHPI Americans. Objective measures of physical activity should be used in subsequent studies. While present findings may not be representative of the greater U.S. NHPI population and subgroups, they underscore the need for larger studies to confirm the prevalence of physical activity and F&V intake and examine correlates of these behaviors.

In this group of obese NHPI adults at high risk for chronic disease, the majority of participants were classified as precontemplators regarding physical activity. These are individuals who have no intention to become active. Although most people view exercise as beneficial they may have little personal incentive to act on that belief and may expect the cost of exercise to outweigh any gains.³⁴ Promoting healthy lifestyle behaviors to a high-risk group which has no intention of changing their behavior poses an additional challenge.

Commonly-observed psychosocial and environmental correlates of physical activity did not appear to generalize to this NHPI sample. If subsequent larger studies replicate present results, it suggests culturally-specific factors are driving physical activity. Those culturally-specific variables need to be identified, beginning with qualitative research methods and followed by quantitative assessments. Such studies are needed to provide an empirical basis for culturally-specific interventions. Pacific Islander societies tend to place more emphasis on the family, group, or community as a whole rather than individual autonomy, comfort and self-care.⁵ Food is fundamental to establishing and sustaining social relationships among NHPI, and collectively the dominant views, values and expectations about physical activity create a social and cultural environment that has the potential to either promote or prevent this behaviors.⁵ It seems reasonable that group endeavors which foster family and community support among NHPI might be more beneficial for adopting and sustaining regular physical activity and increased F&V intake, compared to motivational interventions targeting individuals.