Interactions between individual and perceived environmental factors on Latinas’ physical activity

Abstract

Background

Latinas have disproportionately low levels of physical activity (PA) and the ecological correlates of their PA remain unclear. This study aims to test interactions between individual and environmental factors on Latinas’ PA.

Methods

We analyzed baseline data from 436 Latinas participating in a PA randomized controlled trial in San Diego, CA [Fe en Acción/Faith in Action]. Measures included demographics, perceived environment, PA and anthropometrics. Mixed effects models examined interactions between individual and environmental factors on self-reported leisure-time and transportation, and accelerometer-assessed PA.

Results

Significant positive associations were found between neighborhood aesthetics and leisure-time moderate-to-vigorous PA (MVPA) and between having destinations within walking distance from home and transportation PA (P < 0.05). We found significant interactions of income with aesthetics and sidewalk maintenance as well as between weight status and safety from crime. Favorable aesthetics was related to more leisure-time MVPA only among lower income women (odds ratio (OR) = 1.57; 95% confidence interval (CI): 1.18, 2.08); however, higher income women reporting better sidewalk maintenance reported more leisure-time MVPA (OR = 1.51; 95% CI: 1.06, 2.15). Higher perceived safety from crime was positively related to transportation PA only among overweight/obese women.

Conclusions

Subgroup differences should be considered when developing interventions targeting the neighborhood environment to promote Latinas’ PA.

Introduction

Among US adults, women and racial/ethnic minorities are less likely to meet the physical activity (PA) guidelines than men and non-Latino Whites, respectively.¹ This places these groups at increased risk of multiple health problems.² Nationally, Latinos report less leisure-time PA but more walking for work-related transportation compared to non-Latino Whites.^3,4 Racial/ethnic differences have also been observed for total PA as measured by accelerometry, with Latinos being more active than non-Latino Whites and non-Latino Blacks.⁵ Among Latinos, women (Latinas) tend to be less physically active as assessed by self-report and accelerometer measures compared to men.^3,5 The factors influencing these differences in PA are not well understood, but neighborhoods characterized by high poverty and predominantly immigrant residents may have less favorable environmental characteristics that are not conducive to PA (e.g. low perceptions of safety).^6,7

Among the few ecological studies of Latinas’ PA, certain environmental factors have been cited as either barriers (e.g. lack of sidewalks) or facilitators (e.g. favorable aesthetics) of self-reported leisure-time PA^8–10 and total PA.¹¹ Self-reported transportation PA among adults has also been linked with higher number of destinations and public transit.⁹ Not all environmental characteristics have shown consistent associations with PA, however, (e.g. safety from crime¹²), suggesting that environment-PA associations may differ across subpopulations.^13,14 A few studies have tested interactions between individual and environmental factors^13,15,16 but such interactions have rarely been examined among Latinas.¹⁴

The purpose of this study was to test interactions of individual and perceived environmental factors in relation to Latinas’ self-reported PA (leisure-time and transportation) and accelerometer-assessed total PA. Based on findings from a study showing significant interactions between socio-demographics and neighborhood safety in relation to PA among adults, specifically positive associations among the more affluent/advantaged groups,¹³ we expected associations between perceived environmental factors and both self-reported and accelerometer-assessed PA to be positive in all Latinas and stronger in those with more advantaged individual factors (e.g. higher income). Evidence of significant interactions would suggest that the associations between perceived environmental factors and PA may depend on individual characteristics, which has important implications for tailoring multilevel interventions to promote PA.

Methods

Overview of study

This cross-sectional study used baseline data from Fe en Acción [Faith in Action], a two-group randomized trial for PA promotion among churchgoing low-active Latinas in San Diego, CA. The trial was based on the socio-ecological model of health behavior,¹⁵ targeting multiple levels of influence including the individual (e.g. PA behaviors), interpersonal (e.g. social support), organizational (e.g. church support) and environmental (e.g. neighborhood characteristics) factors.

Baseline data for the study were collected between May 2011 and September 2013. Details of the study's inclusion criteria and research protocols, including sampling, recruitment, data collection, and intervention components are described elsewhere.¹⁷ In brief, 436 self-identified Latinas (18–65 years) were recruited from 16 participating Catholic churches in San Diego County. Between 20 and 35 Latinas were recruited per church. Participant eligibility included: attending the church at least 4 times/month, residing within 15 min driving distance to the church, reporting no health condition that would interfere with their ability to be physically active (e.g. pregnancy), and reporting low levels of PA as assessed by two screeners^18,19 (e.g. no or mostly light-intensity PA during leisure-time or on the job). Women who met the eligibility criteria were then screened for their accelerometer-assessed MVPA with those engaging in ≥250 min/week of MVPA being excluded (n = 113). Although the 250 min/week cut-off exceeded the recommendations for aerobic PA (i.e. at least 150 min/week of MVPA), we used a higher MVPA threshold for inclusion because early in recruitment it became evident that the majority of participants exceeded the 150 min/week threshold possibly from transportation- and work-related PA. A higher accelerometer-assessed MVPA cut-off allowed for inclusion of Latinas mostly in need of a PA promotion intervention. Further details of the PA screening procedures are described elsewhere.¹⁷ The San Diego State University Institutional Review Board approved this study.

Baseline measurements

All baseline measurements were administered by trained, bilingual research assistants (RAs). RAs described the study to the participant and obtained written informed consent. Anthropometric measurements (e.g. height and weight) were collected following standard protocols.²⁰ Participants were also asked to wear an accelerometer for 7 consecutive days and complete a survey in their preferred language (English/Spanish). The survey scales were available in Spanish and reviewed by bilingual staff to ensure conceptual and linguistic equivalence to the English version prior to data collection.

Physical activity

Self-reported

Trained RAs administered the Global Physical Activity Questionnaire (GPAQ)²¹ to assess domain-specific PA. Present analyses used leisure-time MVPA (six items) and transportation PA (walking and bicycling to/from destinations) (three items). The GPAQ has been validated among Latinas in San Diego²² and is an acceptable tool for diverse populations owing to its low cost, ease of administration and adaptability.²³ Leisure-time MVPA was highly right-skewed, with a large proportion reporting 0 min, thus we categorized this variable into three levels: none, low (10–119 min/week) and high (≥120 min/week), with a cutoff of 120 based on an approximate median split of the non-zero values of leisure-time MVPA. Transportation PA was also highly right-skewed; thus we dichotomized the variable into none versus any transportation PA (≥10 min/week), a cutoff used in other studies.⁹

Accelerometry

Participants were asked to wear a GT3-X or GT3-X+ Actigraph accelerometer attached to an elastic belt over the right hip for seven consecutive days. These devices objectively measure intensity, frequency and duration of movement in three axes.²⁴ Participants were instructed to remove the device during water activities (e.g. swimming and bathing) and sleep. The devices were initialized to record in 1-s epochs starting at midnight the day the device was delivered. The minimum required wear time was ≥5 valid days (including ≥1 weekend day) with ≥10 valid hours/day of data.¹⁷ Non-wear time was defined as ≥60 consecutive minutes of zero count values. Participants with <5 valid days of wear were asked to re-wear the device for the number of days needed plus an extra day to ensure collection of sufficient data. Physiologically implausible data, i.e. >15 999 counts/min (cpm),²⁵ were removed from the analyses. Accelerometer files were converted to 60-s epoch files and processed using ActiLife software version 6 (ActiGraph, Pensacola, FL) and the Troiano 2008 cutoff points.⁵ Time spent in MVPA was determined by summing each minute where the count met the criterion for moderate activity (2020 cpm) or vigorous activity (5999 cpm). We estimated average MVPA min/week, counting each minute of moderate and vigorous activity, and analyzed these data as a normally distributed continuous variable.

Demographics

Demographic items were based on questions adapted from the 2005 Behavioral Risk Factor Surveillance System (BRFSS) questionnaire²⁶ and included marital status (married or living as married/single or non-partnered), education (less than high school/high school or higher), employment (yes/no), country of birth, monthly household income (less than $2000/$2000 or more) and the number of vehicles and adults living in the household (to compute the number of vehicles per adult as a proxy for vehicle access). Because income was based on ranges, we could not properly calculate participants’ poverty level. The cut-off of $2000/month was chosen based on preliminary analyses that showed the distribution of responses to the income question was highly right-skewed with about 50% of participants reporting a monthly household income less than $2000. The variable was therefore recoded as a binary variable.

Acculturation

The 24-item Bidimensional Acculturation Scale (BAS) for Hispanics²⁷ assessed acculturation in two cultural dimensions (Hispanic and non-Hispanic). Items assessed language use, linguistic proficiency and use of electronic media (TV and radio) in English/Spanish. This scale has high internal consistency (α = 0.85 for the Hispanic dimension and 0.95 for the non-Hispanic dimension; similar to other studies²⁷). Items used a 4-point Likert scale ranging either from ‘almost never’ (1) to ‘almost always’ (4) or ‘very poorly’ (1) to ‘very well’ (4). We calculated an average score for each cultural dimension, with adherence to a dimension defined as a mean score ≥ 2.5.²⁷

Body mass index

Weight and height measurements were taken twice and averaged. Average weight (kg) was divided by the average height squared (m²) to calculate body mass index (BMI). To facilitate interpretation of results from the moderation analyses, we dichotomized BMI into: normal weight (<25 kg/m²) and overweight/obese (≥25 kg/m²). Participants with BMI's < 18.5 kg/m² comprised less than 1% of the sample; thus we grouped them in the ‘normal’ weight category.

Perceived neighborhood environment

We adapted seven items from the abbreviated Neighborhood Environment Walkability Scale (NEWS-A)²⁸ to assess perceived safety from crime, safety from traffic and neighborhood aesthetics as well as two items from the U.S. Determinants of Exercise in Women Phone Survey²⁹ to assess perceived access to destinations within walking distance of home and sidewalk maintenance. To reduce participant burden from answering both scales, we included only those items deemed most relevant to churchgoing Latinas in San Diego, as informed by a previous focus group study.⁸ Items from both scales have demonstrated moderate-to-high test–retest reliability (e.g. ICC = 0.80 for safety from crime³⁰). The items for perceived safety from crime, safety from traffic and neighborhood aesthetics were assessed using a 5-point Likert scale ranging from ‘strongly disagree’ (1) to ‘strongly agree’ (5); these reflect the NEWS-A response options except we included a ‘neutral’ option.

The mean of two items assessed perceived safety from crime in the neighborhood during the day and at night. One item captured perceived presence of heavy traffic along nearby streets. To examine perceived neighborhood aesthetics, the average of 4 items (e.g. presence of trees along the streets) was used. Perceived access to destinations within walking distance of one's home (e.g. businesses) was assessed with 1 item (yes/no). Perceived sidewalk maintenance was assessed with 1 item among those who reported having sidewalks in their immediate neighborhood (‘not at all maintained’ (1) to ‘very well maintained’ (4)). All continuous perceived neighborhood environment scores were standardized and mean-centered for ease of interpretation.

Data analyses

Generalized linear mixed effects or mixed effects models with appropriate error distributions and link functions were used to model each outcome while adjusting for clustering effects of the churches. We performed ordinal regression to obtain odds ratio (OR) and 95% confidence interval (CI) of increasing amounts of self-reported leisure-time MVPA and logistic regression to obtain ORs and 95% CI of reporting engaging in any transportation PA. A linear regression model with a normal error distribution was used to model increasing amounts of accelerometer-assessed MVPA. Models were adjusted for age and vehicle access. For each model, we tested 20 interactions between individual (e.g. income) and perceived environmental (e.g. safety from crime) factors. All statistical analyses were performed in SAS version 9.4 (SAS Institute Inc., Cary, NC).

RESULTS

Sample descriptives

In this sample of 436 Latinas, the mean age was 44.4 ± 9.6 years; the average time living in the USA was 21.0 ± 10.3 years; approximately 91% of the sample were born in Mexico and 32% had high levels of adherence to the non-Hispanic domain (Table 1). Most women were married or living as married (77%), had a household income < $2000/month (58%) and had <high school education (55%). Over 80% were overweight or obese (Table 1).

Table 1.

Descriptive characteristics of Latina women in San Diego, CA (N = 436). Fe en Acción 2011–2013.

Characteristic	% or mean ± SD
Demographics
Mean age in years	44.4 ± 9.6
Mean years living in the USA	21.0 ± 10.3
Mean number of adults in household	3.0 ± 1.3
Mean number of vehicles in the household	2.0 ± 1.0
Born in Mexico, %a	90.8
High levels of adherence to non-Hispanic domain, %b	32.2
Married/living as married, %	77.3
Monthly household income < $2000, %	58.3
Completed less than high school, %	54.8
Employed, %	65.8
Overweight or obese, %c	83.2
Physical activity
Reported leisure-time MVPA, %
None	55.5
Low (10–119 min/week)	20.9
High (≥120 min/week)	23.6
Any reported transportation physical activity  (≥10 min/week), %	32.1
Mean accelerometer-assessed MVPA min/week	103.2 ± 63.8
Perceived environmental factors
Mean safety from traffic score (1–5 point rating)d	3.6 ± 1.3
Mean safety from crime score (1–5 point rating)d	3.8 ± 1.2
Mean neighborhood aesthetics score (1–5 point rating)d	3.1 ± 1.0
Mean sidewalk maintenance score (1–4 point rating)d,e	3.4 ± 0.7
Has access to destinations within walking distance of  home (yes), %	79.8

Notes: MVPA = moderate-to-vigorous physical activity; SD = standard deviation.

^aRemainder include those born in the USA (n = 33) or other foreign country (n = 7).

^bBased on a score of ≥2.5 on the non-Hispanic domain of the Bidimensional Acculturation Scale.

^cOverweight: 25 ≤ BMI < 30 (n = 164); obese: >30 BMI (n = 197).

^dHigher scores indicate more favorable perceptions.

^eOnly asked among those who reported having sidewalks in their immediate neighborhood (n = 391).

Over 50% of participants reported engaging in no leisure-time MVPA and about a third reported engaging in any transportation PA (Table 1). On average, participants engaged in 103.2 ± 63.8 min/week of accelerometer-assessed MVPA (Table 1). About 80% of participants reported having access to destinations within walking distance of their home (Table 1). On average, participants reported favorable scores on the perceived environment variables (e.g. mean of 3.6 on a 5-point scale for safety from traffic) (Table 1).

Association of individual factors with PA

None of the individual factors were significantly associated with self-reported leisure-time MVPA (Table 2). Self-reported transportation PA was negatively associated with greater vehicle access (OR = 0.77; 95% CI: 0.60, 0.98) and having a household income ≥ $2000/month (OR = 0.62; 95% CI: 0.38, 1.00) (Table 2). Accelerometer-assessed MVPA was negatively associated with greater age (β = −7.82; 95% CI: −14.56, −1.08), greater vehicle access (β = −7.39; 95% CI: −13.87, −0.91) and having high levels of adherence to the non-Hispanic domain (β = −20.24; −35.25, −5.23) (Table 2).

Table 2.

Association of individual factors with self-reported leisure-time and transportation, and accelerometer-assessed physical activity among Latinas in San Diego, CA. Fe en Acción 2011–2013.

Individual factor	Leisure-time MVPA (n = 378)	Transportation physical activity (n = 375)	Accelerometer-assessed MVPA (n = 378)
	OR (95% CI)a	OR (95% CI)a	ß (95% CI)a	t-value	P-value
Ageb	0.90 (0.73, 1.11)	0.83 (0.65, 1.06)	−7.82 (−14.56, −1.08)	−2.28	0.02*
Vehicle accessb	1.09 (0.90, 1.34)	0.77 (0.60, 0.98)*	−7.39 (−13.87, −0.91)	−2.24	0.03*
Monthly household income			−4.34 (−17.56, 8.87)	−0.65	0.52
<$2000	1.00	1.00
≥$2000	0.85 (0.56, 1.28)	0.62 (0.38, 1.00)*
Education			0.36 (−13.89, 14.61)	0.05	0.96
<High school	1.00	1.00
≥High school	1.13 (0.72, 1.76)	0.99 (0.59, 1.66)
High levels of adherence to non-Hispanic domain			−20.24 (−35.25, −5.23)	−2.65	0.008*
No	1.00	1.00
Yes	0.97 (0.61, 1.55)	0.67 (0.38, 1.16)
Weight status			6.44 (−10.39, 23.27)	0.75	0.45
Normal weight	1.00	1.00
Overweight/obese	1.14 (0.67, 1.94)	0.64 (0.36, 1.15)

Notes: MVPA = moderate-to-vigorous physical activity.

^aAdjusted for church clustering effects.

^bVariables are grand mean centered and standardized.

*P < 0.05.

Association of perceived environmental factors with PA

A significant positive association was found between perceived neighborhood aesthetics and self-reported leisure-time MVPA (OR = 1.25; 95% CI: 1.01, 1.53) (Table 3). Self-reported transportation PA was positively associated with perceived access to destinations within walking distance of the home (OR = 2.34; 95% CI: 1.11, 4.92) (Table 3). There was a marginally significant positive association between perceived access to destinations within walking distance of one's home and accelerometer-assessed MVPA (β = 15.52; 95% CI: −1.81, 32.86) (Table 3).

Table 3.

Associations of perceived environmental factors with self-reported leisure-time and transportation, and accelerometer-assessed physical activity among Latinas in San Diego, CA. Fe en Acción 2011–2013.

	Leisure-time MVPA (n = 380)	Transportation physical activity (n = 378)	Accelerometer-assessed MVPA (n = 380)
Perceived environmental factora	OR (95% CI)b	OR (95% CI)b	ß (95% CI)b	t-value	P-value
Safety from traffic	0.99 (0.78, 1.26)	0.97 (0.74, 1.27)	−1.98 (−9.49, 5.53)	−0.52	0.60
Safety from crime	1.19 (0.93, 1.51)	1.07 (0.81, 1.41)	4.17 (−3.36, 11.70)	1.09	0.28
Neighborhood aesthetics	1.25 (1.01, 1.53)*	1.11 (0.88, 1.40)	5.28 (−1.17, 11.73)	1.61	0.11
Sidewalk maintenance	0.97 (0.79, 1.20)	0.90 (0.71, 1.14)	−3.32 (−9.93, 3.29)	−0.99	0.32
Access to destinations within walking distance of home			15.52 (−1.81, 32.86)	1.76	0.08
No	1.00	1.00
Yes	0.78 (0.46, 1.35)	2.34 (1.11, 4.92)*

Notes: MVPA = moderate-to-vigorous physical activity.

^aEnvironment variables (except access) are grand mean centered and standardized.

^bAdjusted for church clustering, age, and vehicle access.

*P < 0.05.

Moderators of the association between perceived environmental factors and PA

For self-reported leisure-time MVPA, we found two significant interactions with household income, one with perceived neighborhood aesthetics (P = 0.03) and the second with perceived sidewalk maintenance (P = 0.0005) (Table 4). A significant positive association between perceived neighborhood aesthetics and leisure-time MVPA was found only among those with a lower household income (OR = 1.57; 95% CI: 1.18, 2.08). A significant positive association between better perceived sidewalk maintenance and leisure-time MVPA was found only among those with a higher household income (OR = 1.51; 95% CI: 1.06, 2.15). For self-reported transportation PA, we found a significant interaction between perceived safety from crime and weight status (P = 0.02). Among the overweight/obese women a positive association was observed between perceived safety from crime and transportation PA, although this was not statistically significant (OR = 1.24, 95% CI: 0.91–1.68).

Table 4.

Significant interactions between individual and perceived environmental factors on self-reported leisure-time and transportation PA among Latinas in San Diego, CA. Fe en Acción 2011–2013.

Interaction	Level of individual factor	ORa (95% CI)
Leisure-time MVPA
Monthly household income × perceived aestheticsb,*	<$2000	1.57 (1.18, 2.08)
	≥$2000	0.96 (0.69, 1.34)
Monthly household income × perceived sidewalk maintenanceb,**	<$2000	0.67 (0.50, 0.90)
	≥$2000	1.51 (1.06, 2.15)
Transportation physical activity
Weight status × perceived safety from crimeb,*	Normal- weight	0.59 (0.33, 1.04)
	Overweight/  obese	1.24 (0.91, 1.68)

Notes: MVPA = moderate-to-vigorous physical activity.

^aAdjusted for church clustering, age, and vehicle access.

^bVariable is grand mean centered and standardized.

Significance of interaction: *P < 0.05, **P < 0.001.

Discussion

Main findings of this study

We found significant main effects for perceived environmental factors and interactions between individual and perceived environmental factors in explaining Latinas’ self-reported leisure-time and transportation, but not accelerometer-assessed PA. The only significant positive environmental main effect of leisure-time MVPA was favorably perceived aesthetics; for transportation PA, it was perceived access to destinations near the home. We found significant interactions of household income with perceived neighborhood aesthetics and sidewalk maintenance on leisure-time MVPA. For transportation PA, one significant interaction was found, i.e. between perceived safety from crime and weight status.

What is already known on this topic

Our findings are consistent with prior studies of individual and environmental correlates of domain-specific PA.^8–10,31,32 A review of studies on the correlates of PA noted evidence for associations of self-reported leisure-time MVPA with age (inversely), income, education, and overweight (inversely).³³ Although we found no significant associations between these individual factors and leisure-time MVPA, our results are similar to those of two other studies with Latina samples.^10,32 We also found no relationship with vehicle access. To our knowledge, no other study has examined the relationship between vehicle access and leisure-time MVPA in a Latino sample. However, we hypothesized that among this sample of Latinas, vehicle access would have a stronger effect on necessity-driven behaviors such as transportation PA to get to/from destinations as seen in other studies^34,35 than choice-driven behaviors such as leisure-time PA. We found a significant association between perceived neighborhood aesthetics and leisure-time MVPA, which is supported by another study of Latinas in San Diego and an international study of adults from 12 countries (including Mexico).^8,36 Better neighborhood aesthetics may discourage vandalism, potentially creating a sense of comfort and safety to exercise in the neighborhood. Our findings suggest that multilevel interventions aimed at promoting Latinas’ leisure-time MVPA can consider improving neighborhood aesthetics as a potential environmental target.

For transportation PA, we found a significant negative relationship with higher household income, which is consistent with findings from a national study.³⁷ We also found a negative relationship with vehicle access, which has been reported in several studies of travel behaviors^34,35 but none that we are aware of involving a Latina sample. Further, Latinos tend to live in more mixed use urban areas, thus improving their access to destinations to walk to may promote walking for transportation.^10,38–40

For accelerometer-assessed MVPA, consistent with other studies,^41,42 we found that those with greater age and vehicle access had fewer weekly minutes of accelerometer-assessed MVPA. Because Latinos tend to walk more for transportation than other racial/ethnic groups,⁴³ having greater vehicle access may reduce their transportation PA and thereby accelerometer-assessed MVPA. We also found a negative relationship between acculturation and accelerometer-assessed MVPA; however, the only other study that we are aware of that has examined the relationship between language-based acculturation and accelerometer activity counts among Latinos found no relationship.⁴⁴ As reported by Berrigan et al. (2006),⁴⁵ Latinos with higher acculturation levels report higher leisure-time PA but lower transportation PA compared to those of lower acculturation. In the process of becoming acculturated to the US, Latinos may improve their socio-economic status and buy cars, which can have negative consequences on their overall PA by reductions in their transportation PA. Regarding the lack of perceived environmental correlates for accelerometer-assessed MVPA, our findings are partly inconsistent with prior studies.^28,30,46 Because the accelerometer captures activity in all domains, this measure may be less likely to correlate with pedestrian/transportation neighborhood environmental factors that are likely to be specifically related to leisure-time or transportation PA. Examination of conceptually mismatched associations (e.g. transportation environment with total PA) may lead to type 2 error¹⁴ and as such, our lack of findings with the accelerometer-assessed MVPA may be a result of weak conceptual matching.

What this study adds

To our knowledge, this is one of the first studies to examine individual level moderators of associations of perceived neighborhood environmental factors with self-reported domain-specific and accelerometer-assessed PA in a Latina sample. As such, our findings are novel and relevant to the development of public health multi-level interventions aimed at promoting Latinas’ PA.

For self-reported leisure-time MVPA, we found significant interactions of household income with perceived neighborhood aesthetics and sidewalk maintenance; however, the positive associations were not always in the expected income group (i.e. higher income women). Higher leisure-time MVPA was significantly related to favorable perceived neighborhood aesthetics among lower income women, but significantly related to favorable perceived sidewalk maintenance among higher income women. The reasons for the inconsistent associations are unknown but one national study found that lower income women report less enjoyable scenery in their neighborhood than higher income women and that enjoyable scenery may be a particularly important environmental correlate of meeting PA recommendations among lower income men and women.⁴⁷ Latinas in our study may have been less likely to engage in choice-driven behaviors such as leisure-time PA. A study from Mexico reported that PA among Mexican adults may be more strongly driven by necessity (transportation PA) than choice (leisure-time PA).⁴¹ Improvements to neighborhood aesthetics may be an important target for promoting leisure-time MVPA among lower income Latinas; however, additional studies are needed to replicate our findings using objective measures of the environment and to examine the influence of other cultural and contextual factors on this relationship.

The significant interaction between household income and perceived sidewalk maintenance indicates that sidewalk maintenance may be a less important correlate of lower income Latinas’ leisure-time MVPA. Lower income adults have been found to report less walking for leisure-time than those of higher income.⁴⁸ However, for the observed interaction and that between income and perceived neighborhood aesthetics, the findings require careful interpretation. Individuals spending more time walking in their neighborhood may also be more aware of their neighborhood features (e.g. sidewalk conditions) than those who are not walking in their neighborhood, e.g. Latinas with higher vehicle access choosing to drive rather than walk for transportation. This potential bias has been reported in another study.⁴⁹ Greater exposure to and thereby awareness of one's neighborhood features may obscure the influence of income on the relation between perceived environment and leisure-time MVPA.

For self-reported transportation PA, we found a significant interaction between weight status and perceived safety from crime. There was a non-significant positive association between increasing levels of safety from crime and reporting any transportation PA among overweight/obese women compared to normal-weight women. To date, the evidence on the relationship between perceived neighborhood safety and PA has shown inconsistent results,¹² pointing to the potential influence of third variables such as weight status acting as either moderators or mediators of this association. Longitudinal studies are needed to better understand the influence of weight status on perceptions of safety from crime and PA.

Limitations of this study

The cross-sectional design of this study did not allow evaluation of a cause-and-effect relationship between the perceived environment and PA outcomes. Self-report evaluation of the neighborhood environment may be biased, and only portions of full measures were used. Our study's focus on Latinas living in a US–Mexico border community limits the generalization of our findings. Further, our research involved low active Latinas, limiting the generalization of the findings to the general Latina population.

Strengths of this study

Among study strengths, we used baseline data from a study that targeted multiple levels of influence of PA behaviors, allowing for an ecological study of interactions of individual and environmental level factors in relation to PA. A key strength was the examination of both self-report domain-specific and accelerometer-assessed total PA outcomes, which allowed for a more comprehensive investigation of Latinas’ PA patterns, correlates and moderators.

Conclusion

This study provides evidence of interactions between multilevel correlates of PA, which supports a principle of ecological models and may lead to place- and person-tailored interventions. Overall, our findings suggest the variability in the associations of perceived environmental factors with self-reported leisure-time and transportation PA may depend on individual factors. Further studies are needed to validate present findings, evaluate their generalizability to other geographical areas, and explore additional interactions across levels of the ecological model. Our findings may help inform interventions targeting multilevel correlates to promote Latinas’ PA.

Funding

This work was supported by the National Cancer Institute of the National Institutes of Health (R01CA138894) and a diversity supplement from the National Cancer Institute (3R01CA138894-04S1).