Coaction Between Physical Activity and Fruit and Vegetable Intake in Racially Diverse,Obese Adults

Abstract

Purpose:

There is minimal understanding of the potential for coaction, defined as action on one behavior increasing the likelihood of taking action on another behavior, between physical activity (PA) and fruit and vegetable (FV) intake. The purpose of this study was to assess the bidirectional coaction between FV intake and PA, as well as self-efficacy for these behaviors, in a racially diverse sample of obese adults.

Design:

This is a secondary analysis using data collected from the Path to Health study, a randomized controlled trial. ClinicalTrials.gov Identifier: NCT03674229.

Sample:

Obese adults who completed baseline and 6-month follow-up assessments.

Measures:

For this study, data on FV intake, leisure time PA, and 7-day accelerometer data were analyzed at baseline and 6-month follow-up.

Analysis:

We interchanged modeling the FV intake and PA change variables as the independent and dependent variables. We conducted multiple imputation and both linear and multinomial regression.

Results:

The sample (n = 168) was 59% female and mainly split between white (42%) and African American (42%). Change in self-efficacy for PA was predictive of change in self-efficacy for FV intake and vice versa. When compared with participants with no change in FV intake, someone with a positive change in FV intake was more likely to have a positive change in self-reported PA (adjusted risk ratio [RR] = 6.72, 95% confidence interval [CI] = 1.69-26.68). Likewise, when compared with no change, participants with a positive change in self-reported PA were more likely to report a positive change in FV intake (adjusted RR = 6.79, 95% CI = 1.70-27.17).

Conclusion:

Findings suggest coaction between self-efficacy for FV intake and PA as well as between FV intake and PA. Coaction could be capitalized on to more effectively promote both energy-balance behaviors.

Keywords

physical activity nutrition weight control interventions coaction

Purpose

Obesity prevalence continues to increase in the United States.^1,2 In addition to the need to continue with prevention efforts, this trend indicates that there will be even more individuals in future years in need of obesity treatment and weight management programs (WMPs). Initially, WMPs focused on dietary behavior, but given the evidence that the inclusion of physical activity (PA) behavior change alongside dietary changes improves weight loss,^3

-7 WMPs now tend to include both behaviors.

Previous literature indicates that there can be coaction⁸ between behaviors, with variations in the concept called carryover,⁹ covariation,¹⁰ spillover,¹¹ or transfer.¹² Coaction is defined as behavior change for one behavior resulting in change in another behavior.^8,13,14 Coaction between PA and dietary behaviors may produce greater change than targeting each behavior separately, which is especially relevant to WMPs.¹³ A better understanding of how the 2 behaviors influence each other can guide decisions on curriculum content and ordering of content in WMPs to achieve better outcomes.^14,15

For fruit and vegetable (FV) intake, just one facet of dietary behavior, it is hypothesized that it shares contexts, regulatory resources, and skills with PA that allow change in one behavior to facilitate or quick-start change in the other.^16
-18 Achieving positive change in PA and FV intake typically requires the addition of something new (eg, unfamiliar produce and additional activities) as opposed to modification or elimination. Although coaction between these 2 behaviors has been investigated using cross-sectional and longitudinal studies,^19

-23 only recently has coaction been studied within the context of multiple behavior change (MBC) interventions. In particular, the results of stage-based models of behavior change have found that participants who are able to progress from action to maintenance of one behavior (healthy eating or PA) are more likely to progress from action to maintenance for the other behavior.^8,17,24

The few studies that have found coaction between 2 behaviors outside the scope of the Transtheoretical Model or other stage-based MBC interventions have focused on how PA impacts diet^25,26 or the theoretical determinants of diet, such as motivation,¹¹ self-regulation,²⁶ and self-efficacy.²⁶ The positive influence of PA on healthy eating and subsequent weight loss has been hypothesized to be because PA helps to control binge eating, improve body image, and enhance mood and affect.^27
-29 However, this relationship is not true for all individuals, as positive change in one behavior can lead to negative change in the other. For example, compensatory eating may occur after PA.^30,31 There have been few studies within the context of WMPs investigating how healthy eating can influence PA.^9,25 There is a need to further explore coaction in both directions during WMPs with obese individuals, especially with more rigorous assessments of these behaviors.¹⁶

Self-efficacy,^32,33 or the confidence to perform a specific behavior, is one of the strongest predictors of both PA and FV intake.^34
-36 Although initially only part of the social cognitive theory (SCT), the ability of self-efficacy to predict behavior³⁷ has since led to its incorporation into many behavioral theories. Despite Bandura’s original definitions of self-efficacy as being highly contextual and behavior- and domain-specific,^32,33,38 the possibility of self-efficacy covarying between behaviors with similar contexts was subsequently acknowledged by Bandura.³⁸ Thus, coaction between self-efficacy for PA and self-efficacy for healthy eating is also possible, as there may be facilitating patterns between the psychological constructs (ie, self-efficacy) that are thought to precede the behaviors.¹⁷ Indeed, some initial studies have shown that within the context of WMPs, self-efficacy developed for PA can be carried over to self-efficacy for healthy eating.^9,17,26,39 Therefore, there is a need to better understand coaction both at the psychological and behavioral levels⁴⁰ so that researchers can design more effective, long-lasting WMPs that address each behavior when it is most appropriate within the course of treatment.

The purpose of this study was to assess the coaction over time between FV intake and PA, as well as between self-efficacy for FV intake and self-efficacy for PA, in a racially diverse group of obese participants using multiple data collection points. We hypothesized that there would be positive coaction between FV intake and PA as well as between self-efficacy for FV intake and self-efficacy for PA.

Methods

Design

This study was a secondary data analysis from the Path to Health study, which was a randomized controlled trial with 168 individuals conducted from 2015 to 2016.⁴¹ Sample size for this study was thus predetermined by a priori power analysis for the parent study. This study was reviewed and approved by appropriate institutional review boards. The goal of Path to Health was to facilitate enrollment to, use of, and adherence to commercially available, evidence-based WMPs, designed to increase PA and improve diet. Path to Health evaluated the efficacy of up to 6 brief calls with a telephone health coach to facilitate enrollment of participants in a WMP. Coaches provided information on features and other aspects of commercially available WMPs to help participants find a program that fit their needs, with some minimal assistance with behavior-specific questions. The health coaches were masters-level trained counselors who used a combination of motivational enhancement and social cognitive treatment techniques to problem-solving derived from motivational interviewing and SCT.⁴² Although we found a positive intervention effect on enrollment in WMPs by 6 months (39% of the intervention group vs 29% of the control group), we did not find an intervention effect on FV intake or PA.⁴¹ Thus, we treated the study as an observational study, and included both intervention and control group participants in these analyses focused on FV intake and PA.

Sample

Participants were patients from Internal Medicine or Family Medicine clinics affiliated with a large academic institution. To be eligible, patients had to be referred by a health-care provider, have a body mass index (BMI) between 30 and 45 kg/m², be ≥18 years of age, able to speak English, have a working telephone number and address where materials could be mailed, able to engage in moderate-intensity PA as determined by the Physical Activity Readiness Questionnaire,⁴³ and have Internet access at home or other community location.

Eligible individuals from participating clinics (n = 1554) were identified by electronic medical record review. A signed letter from their physicians was sent to each potential participant recommending study participation, outlining study details and providing the study phone number to call if interested in learning more. A week after letters were sent, automated calls were made that included similar information. Additional details on recruitment for the study are published elsewhere.⁴¹ Briefly, of the 512 individuals that called the study line, 339 were interested after receiving more details. Of those, 245 were identified as potentially eligible based on preliminary phone screening using previously mentioned study criteria. If an individual was deemed eligible based on initial screening and remained interested in the study, they were scheduled for an in-person baseline visit to confirm their eligibility based on their BMI (see next section for BMI calculation). Many individuals did not schedule an appointment (n = 17) or canceled and no-showed for their appointment (n = 50). Only 5 were ineligible at the in-person screening, 2 decided not to enroll, and 3 immediately withdrew. A total of 168 individuals were enrolled in this parent study.

Data Collection

At the baseline in-person visit, participants completed a written consent. Data were collected on participants at baseline and 6-month follow-up at in-person assessments. Study staff collected height with a stadiometer and weight using the Tanita integrated bioelectrical impedance body fat monitor scale (Tanita Body Fat Analyzer TBF 350; Tanita Corporation of America, Inc, Arlington Heights, Illinois). Participants removed shoes and heavy outer clothing prior to measurements. These 2 measures were used to determine BMI and served as a final screener for inclusion in the study. Participants then completed a self-administered demographic, psychosocial, and behavioral questionnaire using REDCap.⁴⁴ Participants received $40 gift cards.

Measures

Measures collected included basic demographics from the Behavioral Risk Factor Surveillance System (gender, race/ethnicity, and age),⁴⁵ as well as psychosocial and behavioral variables, including self-efficacy, PA, and FV intake.^46

-50 Self-efficacy for FV intake and PA were measured using questions from the Patient-Centered Assessment and Counseling for Exercise psychosocial measures.^46,47 Responses for both self-efficacy scales were on a 5-point scale from “Not at all confident” to “Extremely confident.” Self-efficacy for FV intake included 6 items assessing how confident an individual was that they could eat 5 servings of FV every day, drink 100% fruit juice instead of soda or fruit punch, eat FV for a snack instead of chips or candy, eat FV when eating out at a restaurant, eat FV when upset or having a bad day, and eat FV when at a social event.⁴⁶ The scale had good internal consistency at each time point (Cronbach α at baseline = .81, 6 months = 0.82). Self-efficacy for PA included 5 items assessing how confident an individual was that they could exercise when tired, in a bad mood, feel like they don’t have time, on vacation, or when the weather is bad.⁴⁷ The scale had good internal consistency at each time point (Cronbach α at baseline = .81, 6 months = 0.79).

The National Cancer Institute(NCI)/National Institutes of Health (NIH) FV intake all-day screener, which is comparable to the 24-hour dietary recalls,⁴⁸ is a 19-item screener used to capture FV intake. Items include questions on frequency (per day, week, or month) and usual quantity consumed of various fruits and vegetables at each occasion. Using established protocols,⁴⁸ responses are used to calculate average daily FV servings. Participants wore an Actigraph (ActiGraph wGT3X+, ActiGraph, LLC, Pensacola, FL, USA) accelerometer for 7 consecutive days during waking hours, starting on the first full day after the visit. They completed an accompanying log, indicating what time they put the accelerometer on in the morning and what time they took it off at night, as well as any other instances where it may have been removed. After 7 days, participants mailed back the accelerometers with a prepaid envelope given to them at their in-person visit. Study staff extracted data from the accelerometer using ActiLife version 6/11.8 software. To be included in the analyses, participants had to have 10 or more hours of valid wear time each day for at least 3 nonconsecutive days (2 weekdays and 1 weekend) with bouts of activity defined as 8 of 10 minutes.⁵¹ Any time interval with 60 or more minutes of continuous zero counts, allowing for a 1- to 2-minute interruption, was categorized as nonwear time.⁵² We used established cut points for adults, with 2020 to 5998 counts per minute categorized as moderate intensity and 5999 counts per minute and above categorized as vigorous intensity⁵² to generate the average minutes of moderate-to-vigorous intensity PA per week. The Godin-Shephard Leisure-Time PA Questionnaire, modified to include duration, captured self-reported leisure-time PA with 7 items.^49,50 Responses for moderate and vigorous activity were used, with times per week and duration multiplied and then aggregated to produce the total minutes per week of moderate-to-vigorous intensity PA.

Analysis

Prior to analyses, we removed extreme values from the analyses, including observations of more than 840 minutes of self-reported (n = 3) or accelerometer-measured (n = 2) PA a week and more than 14 servings of FV a day (n = 4). For analyses, we generated continuous change scores by subtracting the baseline values from the values at the 6-month follow-up. We used linear regression to generate unadjusted estimates as well as estimates that controlled for covariates, including gender, age, and race/ethnicity. In addition to the continuous variables, we generated a 3-category variable to indicate whether, over time, an individual (1) decreased, (2) stayed the same, or (3) increased their self-efficacy for FV intake, self-efficacy for PA, and PA or FV intake. As we did not expect individuals to maintain the same score even if they had not truly altered their behavior, we allowed a tolerance in either direction (an interval within which people were allowed to fluctuate), with 20 minutes per week for both accelerometer-measured and self-reported PA, ¼ cup for FV intake per day, and 0.25 units for both self-efficacy scales. The latter value was picked because it represented a shift in response to at least one of the self-efficacy questions within the larger scale. For these categorical variables (decreased, stayed the same, increased), we conducted multinomial logistic regression controlling for gender, age, and race/ethnicity. For all analyses, we first modeled PA as the independent variable and FV intake as the dependent variable. We then performed the reverse, with FV intake treated as the independent variable and PA treated as the dependent variable. Given the amount of missing data primarily due to lost to follow-up (n = 42 for both change in self-efficacy for FV intake and change in self-efficacy for PA; n = 48 for change in FV intake, n = 45 for change in self-reported PA, and n = 72 for change in accelerometer PA),we conducted multiple imputation with 50 iterations prior to analyses. All analyses were conducted using Stata/SE 14.2 (College Station, Texas).

Results

The sample was 59% women, with the majority of participants having a 4-year college degree (55%) and were married or living with a partner (53%; Table 1). The average age was 58, and average BMI was 36 kg/m² (range from 28.9 to 47.4). The sample was racially diverse, with 42% white, 42% African American, 11% Hispanic, and 5% other.

Table 1.

Demographic Characteristics of Obese Adult Participants in Houston.^a

Variables	N = 168	%
Age, mean (SD)	57.5	12.9
BMI, mean (SD)	36.0	4.3
Gender
Male	69	41.1
Female	99	58.9
Race/ethnicity
Hispanic	19	11.3
White	71	42.3
Black	70	41.7
Other	8	4.7
Education
Grades 9 through 12 (some high school)	2	1.2
High school graduate or GED	16	9.5
College 1 year to 3 years (some college or technical school)	58	34.5
College 4 years or more (college graduate)	92	54.8
Marital status
Married or living with a partner	89	53.0
Divorced, separated, or widowed	42	25.0
Never married	36	21.4
Refused to answer	1	0.6

Abbreviations: BMI, body mass index (kg/m²); SD, standard deviation.

^a N = 168.

Linear Regression

Summary statistics for the linear change scores are presented in Table 2. The medians for changes in FV intake, self-reported PA, and accelerometer-measured PA were all close to 0. On average, most individuals had a slight decrease in their self-reported self-efficacy at the 6-month follow-up compared to baseline.

Table 2.

Descriptive Statistics for Continuous Change Scores.

Variable	N	Median	Mean	SD	Range
Variable	N	Median	Mean	SD	Min	Max
ΔSelf-efficacy for PA	126	−0.40	−0.35	0.86	−3.00	2.00
ΔSelf-efficacy for FV intake	126	−0.17	−0.22	0.77	−3.00	2.67
ΔFV intake	120	0.05	0.03	2.43	−7.12	7.02
ΔSelf-reported PA	123	0	19.08	150.70	−461.00	625.00
ΔAccelerometer PA	96	−0.05	−33.85	142.15	−498.63	369.50

Abbreviations: FV, fruit and vegetable; PA, physical activity; SD, standard deviation; Δ, difference between 6-month follow-up score and baseline score.

Results indicate that in both unadjusted and adjusted models, the change in self-efficacy for PA for the 6-month period was significantly associated with change in self-efficacy for FV intake over the same 6-month period (Table 3). Similarly, change in self-efficacy for FV intake for the 6-month period was significantly associated with change in self-efficacy for PA over the same 6-month period, indicating coaction. There was no coaction between PA and FV intake.

Table 3.

Regression Coefficients for Linear Regression Models Assessing Coaction Between Behaviors and Types of Self-Efficacy in an Obese Sample of Houston Adults.

Independent	Dependent	Unadjusted B	95% CI	Adjusted^a B	95% CI
ΔSelf-efficacy for PA	ΔSelf-efficacy for FV intake	0.22^b	0.07 to 0.36	0.20^c	0.05 to 0.35
ΔSelf-efficacy for FV intake	ΔSelf-efficacy for PA	0.27^b	0.08 to 0.45	0.25^c	0.05 to 0.45
ΔSelf-reported PA	ΔFV intake	0.00	0 to 0	0.00	0 to 0
ΔFV intake	ΔSelf-reported PA	3.03	−9.64 to 15.69	3.28	−9.37 to 14.93
ΔAccelerometer PA	ΔFV intake	0.00	0 to 0	0.00	0 to 0
ΔFV intake	ΔAccelerometer PA	4.75	−7.50 to 17.00	3.26	−8.65 to 15.17

Abbreviations: CI, confidence interval; FV, fruit and vegetable; PA, physical activity; Δ, difference between 6-month follow-up score and baseline score.

^a Models controlled for gender, age, and race/ethnicity.

^b P < .01.

^c P < .05.

Multinomial Logistic Regression

We also examined the behaviors and self-efficacy as categorical variables, with catgories labeled positive, negative, or no change. About 19% and 14% of the sample had a positive change in self-efficacy for FV intake and self-efficacy for PA, respectively (data not shown). About 28% of the sample increased their FV intake by more than ¼ cup daily. Lastly, about 18% and 30% of the sample increased their accelerometer-measured and self-reported PA, respectively, by more than 20 minutes a week.

When compared with participants who experienced no change in self-reported PA over the 6-month period, someone with a positive change in self-reported PA was nearly 7 times more likely (P < .01) to have a positive change in FV intake (Table 4). Similarly, compared to individuals who experienced no change in FV intake over the 6-month period, someone who experienced a positive change in FV intake was about 7 times more likely (P < .01) to have a positive change in self-reported PA. In the multinomial analyses, coaction was not identified between accelerometer-measured PA and FV intake or between self-efficacy for FV intake and self-efficacy for PA.

Table 4.

Relative Risks for Multinomial Logistic Regression Models Assessing Coaction Between Behaviors and Self-Efficacy.

Independent	Dependent	Unadjusted RR	95% CI	Adjusted RR ^a	95% CI
ΔSelf-efficacy for PA	Ref: No change
−ΔSelf-efficacy for PA	−ΔSelf-efficacy for FV intake	0.89	0.33-2.41	1.11	0.41-2.99
−ΔSelf-efficacy for PA	+ΔSelf-efficacy for FV intake	0.43	0.11-1.69	0.53	0.13-2.26
+ΔSelf-efficacy for PA	−ΔSelf-efficacy for FV intake	0.80	0.24-2.66	0.89	0.28-2.82
+ΔSelf-efficacy for PA	+ΔSelf-efficacy for FV intake	1.51	0.38-6.02	1.80	0.47-6.87
ΔSelf-efficacy for FV intake	Ref: No change
−ΔSelf-efficacy for FV intake	−ΔSelf-efficacy for PA	0.89	0.33-2.41	1.10	0.41-2.97
−ΔSelf-efficacy for FV intake	+ΔSelf-efficacy for PA	0.80	0.24-2.66	0.89	0.28-2.81
+ΔSelf-efficacy for FV intake	−ΔSelf-efficacy for PA	0.43	0.11-1.69	0.52	0.12-2.21
+ΔSelf-efficacy for FV intake	+ΔSelf-efficacy for PA	1.51	0.38-6.02	1.78	0.47-6.75
ΔSelf-reported PA	Ref: No change
−ΔSelf-reported PA	−ΔFV intake	1.93	0.61-6.14	2.11	0.64-6.94
−ΔSelf-reported PA	+ΔFV intake	2.97	0.82-10.70	3.47	0.93-12.98
+ΔSelf-reported PA	−ΔFV intake	2.01	0.55-7.30	2.32	0.63-8.59
+ΔSelf-reported PA	+ΔFV intake	5.23^b	1.32-20.73	6.79^c	1.70-27.17
ΔFV intake	Ref: No change
−ΔFV intake	−ΔSelf-reported PA	1.93	0.61-6.14	2.18	0.66-7.21
−ΔFV intake	+ΔSelf-reported PA	2.01	0.55-7.30	2.36	0.64-8.72
+ΔFV intake	−ΔSelf-reported PA	2.97	0.82-10.70	3.48	0.93-12.98
+ΔFV intake	+ΔSelf-reported PA	5.23^b	1.32-20.73	6.72^c	1.69-26.68
ΔAccelerometer PA	Ref: No change
−ΔAccelerometer PA	−ΔFV intake	0.40	0.07-2.32	0.38	0.06-2.57
−ΔAccelerometer PA	+ΔFV intake	0.17	0.03-1.02	0.15	0.02-1.07
+ΔAccelerometer PA	−ΔFV intake	0.30	0.05-1.79	0.33	0.05-2.22
+ΔAccelerometer PA	+ΔFV intake	0.23	0.04-1.32	0.26	0.04-1.74
ΔFV intake	Ref: No change
−ΔFV intake	−ΔAccelerometer PA	0.40	0.07-2.32	0.41	0.06-2.71
−ΔFV intake	+ΔAccelerometer PA	0.30	0.05-1.79	0.40	0.06-2.26
+ΔFV intake	−ΔAccelerometer PA	0.17	0.03-1.02	0.17	0.02-1.18
+ΔFV intake	+ΔAccelerometer PA	0.23	0.04-1.32	0.28	0.04-1.86

Abbreviations: CI, confidence interval; FV, fruit and vegetable; PA, physical activity; Ref, reference group; RR, risk ratio; Δ, difference between 6-month follow-up score and baseline score.

^a Models controlled for gender, age, and race/ethnicity.

^b P < .01.

^c P < .05.

Discussion

In this study, we assessed the potential for coaction between self-efficacy for PA and self-efficacy for FV intake, as well as coaction between FV intake and PA, using both self-reported and accelerometer-measured PA. As hypothesized, in multinomial analyses, there was a significant positive relationship between positive change in self-reported PA and positive change in FV intake over the 6-month period. This finding has been seen previously in intervention studies^8,12,16,24 and extends the work of Prochaska and colleagues by showing that coaction also occurs in MBC interventions that are not based on the Transtheoretical model.⁸ Our results further reveal that within the course of a weight management intervention where there was actually little to no significant intervention effect on PA and FV intake behavior in the sample as a whole, individuals who did make positive change in 1 behavior were more likely to make a positive change on another. Given the burden of obesity in the United States,^1,2 interventions targeting one behavior should also consider including the other behavior to capitalize on coaction and help adults achieve better health outcomes. For WMPs in particular, the presence of coaction should be monitored, and additional intervention content could explicitly draw participants’ attention to the potential to transfer acquired self-efficacy, skills learned, and/or processes used for one behavior to the other.¹⁸ Although this may enhance the overall success of WMPs, testing of this type of explicit intervention content highlighting coaction versus standard WMPs content is needed.

We also found that there was a direct association between self-efficacy for PA and self-efficacy for FV intake over time. This finding is in line with the common factor approach proposed by Prochaska¹⁴ and the limited research that has been conducted in this area.^9,17,26,39 As demonstrated by Miao and colleagues, as self-efficacy for one energy-balance behavior increases, a more general sense of efficacy to deal with obstacles and barriers increases and likely promotes transfer of self-efficacy for the other energy-balance behavior.⁹ In 2 separate studies by Annesi and colleagues that assessed the association of self-efficacy for PA with self-efficacy for FV intake in a group of obese adults, they found a significant positive relationship.^26,39 A positive relationship between the self-efficacy variables was not present in our multinomial analyses, which assessed the association of positive–positive and negative–negative changes, as opposed to raw change scores. This finding may be due to a reduction in power caused by the increase in parameters to be estimated in the multinomial models. It is also important to note that in the overall intervention, self-efficacy actually decreased slightly; this may point to the spiral of difficulties that many individuals may face in weight loss attempts. The directionality of self-efficacy for these 2 energy-balance behaviors needs to be explored within the context of MBC programs, especially the potential effect of self-efficacy for FV intake on self-efficacy for PA.

This study has several null findings that differ from results from previous studies. Negative change in self-reported PA was not associated with negative change in FV intake over the 6-month period. This phenomenon has been rarely studied, although epidemiology studies have shown clustering of unhealthy behaviors.^53
-55 There were no associations between accelerometer-measured PA and self-reported FV intake in either linear or multinomial regression analyses. Accelerometer-measured PA is considered more objective than self-reported PA, the latter of which may often produce bias,⁵⁶ since many people fail to accurately judge their lifestyle behaviors. It is thus conceivable that self-reported measures of PA and FV intake are more closely associated with one another and that accelerometer-measured PA would be associated with more objective measures of dietary intake that limit recall and response bias. In our results, only changes in self-reported behaviors were associated with one another, that is, perceived change on one behavior was associated with perceived change on the other behavior. However, this may be advantageous for achieving eventual coaction between the actual energy-balance behaviors. Lastly, the relation between a positive change in self-reported PA and a negative change in FV intake was not significant, which is in contrast to existing literature on compensatory eating of more calorie-dense and less nutrient-dense foods (ie, fruits and vegetables) occurring with increases in PA.^30,31

This study had several limitations. We acknowledge that the original study was not powered for these specific analyses. However, we conducted sensitivity analyses with complete cases only, which confirmed the same overall findings. Additionally, our sample was particularly well-educated, limiting the generalizability of results to the general population. Future research is needed with samples that represent a wider variety of socioeconomic statues, but that retain ethnic/racial diversity. Lastly, although the NCI/NIH FV intake screener has been used in a wide variety of intervention studies, a stronger measure such as the ASA24 hour dietary recall may have been more sensitive to individual change over time. However, this study used strong validated measures of PA and self-efficacy, including accelerometer-measured PA, that were assessed over time within the course of an intervention aiming for weight loss or management. We also conducted sensitivity analyses with complete cases, which produced the same overall findings.

Future research should continue to assess coaction between nutrition and PA and its mechanism(s) in obese populations, as both behaviors are important for weight loss and maintenance.^6,57 This research could use larger samples to identify moderators and potential subgroups of individuals for whom coaction is particularly pronounced. Lastly, we did not investigate the timing for the delivery of the FV intake and PA content, so future research could explicitly test the timing of the delivery, as it is still unclear whether simultaneous or sequential interventions are superior.¹⁵

Conclusions

Study results indicate coaction between self-efficacy for FV intake and self-efficacy for PA as well as between the 2 energy-balance behaviors. The coaction between these behaviors could be capitalized on within the course of an intervention to efficiently and effectively improve both behaviors, ultimately increasing weight loss and health outcomes.

So What?

What is already known on this topic?

There is some evidence for the presence of coaction (action on one behavior increasing the likelihood of taking action on another behavior) between fruit and vegetable (FV) intake and physical activity (PA).

What does this article add?

In alignment with previous research, the findings suggest the presence of coaction between FV intake and self-reported PA in a sample of obese adults. We also found that coaction extended to self-efficacy; change in self-efficacy for one behavior was predictive of change in self-efficacy for the other behavior.

What are the implications for health promotion practice or research?

Coaction could be capitalized on to more effectively promote both of these behaviors. For weight management programs (WMPs) in particular, the presence of coaction should be monitored, and additional intervention content could explicitly draw participants’ attention to the potential to transfer acquired self-efficacy or other skills from one behavior to the other.

Footnotes

Authors’ Note

Maria E. Fernandez is also affiliated with School of Public Health, Houston, TX, USA; Center for Health Promotion and Prevention Research, UTHealth School of Public Health, Houston, TX, USA. Alexandra E. van den Berg and Casey P. Durand are also affiliated with Michael & Susan Dell Center for Healthy Living, UTHealth School of Public Health, Austin, TX, USA. Harold W. Kohl III, is now affiliated with Michael & Susan Dell Center for Healthy Living, UTHealth School of Public Health, Austin, TX, USA; Department of Epidemiology, Human Genetics and Environmental Sciences, UTHealth School of Public Health, Houston, TX, USA; Department of Kinesiology and Health Education, The University of Texas at Austin, Austin, TX, USA. Belinda M. Reininger is now affiliated with Department of Health Promotion and Behavioral Sciences, UTHealth School of Public Health in Brownsville, Brownsville, TX, USA.

Acknowledgments

The authors thank Gary Deyter for editorial assistance.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by National Cancer Institute(NCI)/National Institutes of Health (NIH) through the UTHealth School of Public Health Cancer Education and Career Development Program (R25CA57712) and The University of Texas MD Anderson’s Cancer Center Support Grant (CA016672), by the National Center for Advancing Translational Sciences (UL1 TR000371), by the Cancer Prevention and Research Institute of Texas (RP170259) and by the Duncan Family Institute through the Center for Community-Engaged Translational Research.

ORCID iD

Natalia I. Heredia, PhD, MPH

References

Flegal

Kruszon-Moran

Carroll

Fryar

Ogden

. Trends in obesity among adults in the united states, 2005 to 2014. JAMA. 2016;315(21):2284–2291.

Ogden

Carroll

Fryar

Flegal

. Prevalence of obesity among adults and youth: united states, 2011–2014. NCHS Data Brief. 2015;219(219):1–8.

Greaves

Sheppard

Abraham

, et al. Systematic review of reviews of intervention components associated with increased effectiveness in dietary and physical activity interventions. BMC Public Health. 2011;11(1):119.

Miller

Koceja

Hamilton

. A meta-analysis of the past 25 years of weight loss research using diet, exercise or diet plus exercise intervention. Int J Obes. 1997;21(10):941–947.

Gao

Chen

Van Dam

. Long-term effectiveness of diet-plus-exercise interventions vs. diet-only interventions for weight loss: a meta-analysis. Obes Rev. 2009;10(3):313–323.

Curioni

Lourenco

. Long-term weight loss after diet and exercise: a systematic review. Int J Obes. 2005;29(10):1168–1174.

Johns

Boyce

Jebb

Aveyard

. Behavioural Weight Management Review Group. Diet or exercise interventions vs combined behavioral weight management programs: a systematic review and meta-analysis of direct comparisons. J Acad Nutr Diet. 2014;114(10):1557–1568.

Johnson

Paiva

Mauriello

Prochaska

Redding

Velicer

. Coaction in multiple behavior change interventions: consistency across multiple studies on weight management and obesity prevention. Health Psychol. 2014;33(5):475.

Miao

Gan

Zhou

. Carry-over effect between diet and physical activity: the bottom-up and top-down hypotheses of hierarchical self-efficacy. Psychol Health Med. 2017;22(3):266–274.

10.

Lawler

Winkler

Reeves

Owen

Graves

Eakin

. Multiple health behavior changes and co-variation in a telephone counseling trial. Ann Behav Med. 2010;39(3):250–257.

11.

Mata

Silva

Vieira

, et al. Motivational “spill-over” during weight control: increased self-determination and exercise intrinsic motivation predict eating self-regulation. Health Psychol. 2009;28(6):709–716.

12.

Fleig

Kerschreiter

Schwarzer

Pomp

Lippke

. ‘Sticking to a healthy diet is easier for me when I exercise regularly’: cognitive transfer between physical exercise and healthy nutrition. Psychol Health. 2014;29(12):1361–1372.

13.

Yin

Prochaska

Rossi

, et al. Treatment-enhanced paired action contributes substantially to change across multiple health behaviors: secondary analyses of five randomized trials. Transl Behav Med. 2013;3(1):62–71.

14.

Prochaska

. Multiple health behavior research represents the future of preventive medicine. Prev Med. 2008;46(3):281–285.

15.

Hyman

Pavlik

Taylor

Goodrick

Moye

. Simultaneous vs sequential counseling for multiple behavior change. Arch Intern Med. 2007;167(11):1152–1158.

16.

Annesi

Porter

. Reciprocal effects of treatment-induced increases in exercise and improved eating, and their psychosocial correlates, in obese adults seeking weight loss: a field-based trial. Int J Behav Nut Phys Activ. 2013;10(1):133.

17.

Fleig

Küper

Lippke

Schwarzer

Wiedemann

. Cross-behavior associations and multiple health behavior change: a longitudinal study on physical activity and fruit and vegetable intake. J Health Psychol. 2015;20(5):525–534.

18.

Lippke

. Modelling and supporting complex behavior change related to obesity and diabetes prevention and management with the compensatory carry-over action model. J Diabet Obes. 2014;1(11):1–5.

19.

Vries

Kremers

Smeets

Reubsaet

. Clustering of diet, physical activity and smoking and a general willingness to change. Psychol Health. 2008;23(3):265–278.

20.

Simoes

Byers

Coates

Serdula

Mokdad

Heath

. The association between leisure-time physical activity and dietary fat in American adults. Am J Public Health. 1995;85(2):240–244.

21.

Gillman

Pinto

Tennstedt

Glanz

Marcus

Friedman

. Relationships of physical activity with dietary behaviors among adults. Prev Med. 2001;32(3):295–301.

22.

Berrigan

Dodd

Troiano

Krebs-Smith

Barbash

. Patterns of health behavior in US adults. Prev Med. 2003;36(5):615–623.

23.

Shuval

Nguyen

Yaroch

Drope

Gabriel

. Accelerometer determined sedentary behavior and dietary quality among US adults. Prev Med. 2015;78:38–43.

24.

Johnson

Paiva

Cummins

, et al. Transtheoretical model-based multiple behavior intervention for weight management: effectiveness on a population basis. Prev Med. 2008;46(3):238–246.

25.

Annesi

Mareno

. Temporal aspects of psychosocial predictors of increased fruit and vegetable intake in adults with severe obesity: mediation by physical activity. J Community Health. 2014;39(3):454–463.

26.

Annesi

Marti

. Path analysis of exercise treatment-induced changes in psychological factors leading to weight loss. Psychol Health. 2011;26(8):1081–1098.

27.

Andrade

Coutinho

Silva

, et al. The effect of physical activity on weight loss is mediated by eating self-regulation. Patient Educ Couns. 2010;79(3):320–326.

28.

Baker

Brownell

. Physical activity and maintenance of weight loss: Physiological and psychological mechanisms. In: Christopher

, ed. Physical Activity and Obesity. Champaign, IL: Human Kinetics; 2000:311–328.

29.

Joseph

Alonso

Bond

Leone

Blackburn

. The neurocognitive connection between physical activity and eating behaviour. Obes Rev. 2011;12(10):800–812.

30.

Finlayson

Bryant

Blundell

King

. Acute compensatory eating following exercise is associated with implicit hedonic wanting for food. Physiol Behav. 2009;97(1):62–67.

31.

Melanson

Keadle

Donnelly

Braun

King

. Resistance to exercise-induced weight loss: compensatory behavioral adaptations. Med Sci Sports Exerc. 2013;45(8):1600.

32.

Bandura

. Self-efficacy: toward a unifying theory of behavioral change. Psychol Rev. 1977;84(2):191.

33.

Bandura

. Self-efficacy mechanism in human agency. Am Psychol. 1982;37(2):122.

34.

Kreausukon

Gellert

Lippke

Schwarzer

. Planning and self-efficacy can increase fruit and vegetable consumption: a randomized controlled trial. J Behav Med. 2012;35(4):443–451.

35.

Guillaumie

Godin

Vézina-Im

L-A

. Psychosocial determinants of fruit and vegetable intake in adult population: a systematic review. Int J Behav Nutr Phys Act. 2010;7(1):12.

36.

Bauman

Reis

Sallis

, et al.

Correlates of physical activity: why are some people physically active and others not?

Lancet. 2012;380(9838):258–271.

37.

Bandura

. Social cognitive theory of self-regulation. Organ Behav Hum Decis Process. 1991;50(2):248–287.

38.

Bandura

. Guide for Constructing Self-Efficacy Scales Self-efficacy Beliefs of Adolescents: Charlotte, NC. Information Age Publishing; 2006.

39.

Annesi

Johnson

McEwen

. Changes in self-efficacy for exercise and improved nutrition fostered by increased self-regulation among adults with obesity. J Prim Prevent. 2015;36(5):311–321.

40.

Geller

Lippke

Nigg

. Future directions of multiple behavior change research. J Behav Med. 2017;40(1):194–202.

41.

Heredia

Lee

Hwang

Reininger

Fernandez

McNeill

. Health coaching to encourage obese adults to enroll in commercially-available weight management programs: the path to health study. Contemp Clin Trials. 2019;83:1–9.

42.

Vidrine

Reitzel

Figueroa

, et al. Motivation and problem solving (MAPS): motivationally based skills training for treating substance use. Cogn Behav Pract. 2013;20(4):501–516.

43.

Thomas

Reading

Shephard

. Revision of the Physical Activity Readiness Questionnaire (PAR-Q). Can J Sport Sci. 1992;17(4):338–345.

44.

Harris

Taylor

Thielke

Payne

Gonzalez

Conde

. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42(2):377–381.

45.

Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance System Survey Questionnaire. Atlanta, GA: Department of Health and Human Services, Centers for Disease Control and Prevention; 2013.

46.

Norman

Carlson

Sallis

Wagner

Calfas

Patrick

. Reliability and validity of brief psychosocial measures related to dietary behaviors. Int J Behav Nut Phys Activ. 2010;7(1):56.

47.

Carlson

Sallis

Wagner

, et al. Brief physical activity-related psychosocial measures: reliability and construct validity. J Phys Activ Health. 2012;9(8):1178–1186.

48.

Thompson

Subar

Smith

, et al. Fruit and vegetable assessment: performance of 2 new short instruments and a food frequency questionnaire. J Am Diet Assoc. 2002;102(12):1764–1772.

49.

Godin

Shephard

. A simple method to assess exercise behavior in the community. Can J Appl Sport Sci. 1985;10(3):141–146.

50.

Amireault

Godin

. The Godin-Shephard leisure-time physical activity questionnaire: validity evidence supporting its use for classifying healthy adults into active and insufficiently active categories. Percept Mot Skills. 2015;120(2):604–622.

51.

Trost

McIver

Pate

. Conducting accelerometer-based activity assessments in field-based research. Med Sci Sports Exerc. 2005;37(11): S531.

52.

Troiano

Berrigan

Dodd

Masse

Tilert

McDowell

. Physical activity in the united states measured by accelerometer. Med Sci Sports Exerc. 2008;40(1):181–188.

53.

Héroux

Janssen

Lee

Sui

Hebert

Blair

. Clustering of unhealthy behaviors in the aerobics center longitudinal study. Prev Sci. 2012;13(2):183–195.

54.

De Vries

van’t Riet

Spigt

, et al. Clusters of lifestyle behaviors: results from the Dutch smile study. Prev Med. 2008;46(3):203–208.

55.

Noble

Paul

Turon

Oldmeadow

. Which modifiable health risk behaviours are related? a systematic review of the clustering of smoking, nutrition, alcohol and physical activity (‘SNAP’) health risk factors. Prev Med. 2015;81:16–41.

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(1):56.

57.

Dombrowski

Knittle

Avenell

Araujo-Soares

Sniehotta

. Long term maintenance of weight loss with non-surgical interventions in obese adults: systematic review and meta-analyses of randomised controlled trials. BMJ. 2014;348:g2646.