Abstract

Research indicates that approximately 80% of adolescents do not meet the physical activity (PA) recommendations for Americans. Motivational Interviewing (MI) is one method researchers have examined to increase PA. This systematic review aimed to 1) explore the literature on the effects of MI on PA levels among adolescents and 2) report on the current state of the MI and adolescent PA literature for evidence and gaps. A modified Cochrane method of the systematic review was conducted within relevant databases (PsycInfo, MEDLINE, SPORTDiscus, SCI-EXPANDED, and SSCI). The lead author conducted an initial review of the titles and abstracts. The second review tier consisted of three authors independently conducting a full-text review of the remaining articles, discussing to a consensus whether to retain or reject each article based on inclusion/exclusion criteria. Of the 392 articles initially identified, nine studies were retained in the review. Three of the nine studies found that MI had a statistically significant impact on PA behaviors in adolescents. Two studies found that MI significantly impacted Body Mass Index or body composition. Generalizations are limited due to heterogeneity. Despite this, results indicate that MI is a possible pathway to improve PA behavior in adolescents.

1. Introduction

The current physical activity (PA) guidelines for Americans stipulate that children and adolescence (aged 6 - 17 years) should participate in at least 60 minutes of moderate-to-vigorous PA per day ¹, with the majority of this time spent in aerobic PA. Muscle and bone-strengthening exercises should also be included at least three days per week ¹. These guidelines are consistent with those of other countries and the World Health Organization (WHO) ². Unfortunately, in 2017, only 26.7% of high schoolers met the PA guidelines ³. An additional study showed that as few as 20% of adolescents aged 13-15 met the PA guidelines, with boys being more active than girls ⁴. Adolescents, which WHO defines as individuals in the 10-19 year age group, is a critical time for behavior development, as behavior established during adolescence carries over into adulthood ⁵. Studies suggest that PA levels increase in early childhood and then decline through adolescence and adulthood ^{6, 7}. These findings are concerning, considering physical inactivity is a modifiable risk factor in children and adolescents for several chronic diseases, including cardiovascular disease, obesity, and diabetes ^{8, 9}. Thus, pathways to encourage more PA during adolescence are needed. It is further evident that autonomy-supportive environments lead to greater intrinsic motivation ^{10, 11}, which is associated with habitual exercise behavior ^{12, 13}. Researchers and interventionists can create an autonomy-supportive environment by minimizing external incentives, avoiding controlling language, and providing and respecting the choices of the individuals ^{14, 15}. One pathway through which a researcher or interventionist can create an autonomy-supportive environment is Motivational Interviewing (MI).

MI is a person-centered form of communication that fosters change talk and respects the patients’ autonomy ¹⁶. The goal of MI is to let the individual hear themself state the reasons for a behavior change and guide them towards setting their own health behavior change goals. In adults, MI has been found to impact improvements in several health behaviors, including eating habits, PA, and disease management ¹⁷. Emerging studies among children and adolescents have found a similar improvement in health behaviors such as substance use, diet, sexual health, diabetes, and asthma ^{18, 19}. It is further evident that MI can have positive outcomes in young children’s anthropometric measures ²⁰, and may be an effective method for promoting PA in adolescents. ¹⁸

As this age range is critical for lifelong PA habits, and evidence indicates almost 80% of adolescents do not meet the recommendations for Americans, ⁴ it is vital to examine the effects of MI on adolescent PA levels ^{21, 22}. At the time of writing this paper, the authors could not locate a systematic review examining MI's impact on PA levels in adolescence. Thus, this review aims to 1) explore the literature for the effects of MI on PA levels among adolescents and 2) report on the current state of the literature for evidence and gaps for MI as an intervention for PA in adolescents.

2. Methods

A modified Cochrane method ²³ of the systematic review was conducted within relevant databases (PsycInfo, MEDLINE, SPORTDiscus through EBSCO interface, and SCI-EXPANDED and SSCI through Web of Science). Search terms included “motivational interview*” AND “adolescen* OR teen* OR youth*” AND “physical activity OR exercise OR activity level OR sedentary behavior.”

Articles were included based on the following criteria (1) utilized MI as an intentional and named part of the intervention (not just MI-trained practitioners), (2) targeted adolescents aged 10-18 years, and (3) intervention aimed to increase PA or exercise behavior or decrease sedentary behavior and included some form of PA measurement (self-report or accelerometer), (4) published in the English language between years 1990-2021 (because MI applications began to move into the health care realm in the early 1990s), and (5) peer-reviewed. To ensure sufficient data were collected, all studies that met these criteria were eligible regardless of study design. Researchers did exclude reviews, meta-analyses, protocol articles, conference abstracts, and commentaries.

The lead author searched all databases and exported search reports to Endnote X9. Duplicates were automatically removed by Endnote and then manually by the same author. The lead author then conducted an initial review of the titles and abstracts to remove articles that noticeably did not meet the inclusion criteria. The second review tier consisted of three authors independently conducting a full-text review of the remaining articles, discussing to consensus on the decision to retain or reject each article

The lead author developed a data extraction sheet to extract study characteristics, including study design, participant characteristics, training details, and outcome measures. The primary outcome measures of interest were changes in PA. Secondary outcome measures of interest were changes in BMI or body composition, eating habits, and exercise behavior. After full-text review, study details for each of the retained articles were entered into a tailored data extraction sheet by the lead author for organization and to report study characteristics. In addition to a narrative synthesis, researchers used the Cochran Risk of Bias tool to assess the methodological quality of retained articles.

3. Results

The initial search of the databases yielded 392 articles. Figure 1 displays the Preferred Reporting Items of Systematic Reviews and Meta-Analyses (PRISMA) trial flow diagram. This figure describes each search and review tier’s results and reasons for rejection. After removing duplicates, 291 articles were identified for the initial review of titles and abstracts. This led to the exclusion of 261 articles based on the following: language (n=7), non-intervention studies (i.e., reviews, protocol articles, abstracts, and commentaries; n=90), incorrect age group (n=96), not including MI as part of the intervention (n=33), and not having PA measurements as an outcome (n=35). For the next review tier, the full-text review was conducted for 30 articles; 21 were rejected. Seven articles were rejected for not having PA as outcome measurement, 11 were rejected due to the age of study participants, and 3 for incomplete data. Of the three papers with incomplete data, two were protocols and did not include measures; the third paper did not present the pre/post PA data. We retained nine interventions that used MI as a behavior intervention to increase PA. [24-32]

A summary of retained study characteristics and outcomes is provided in Table 1. The mean age of participants ranged from 13 to 17 years old. Samples ranged from 30 to 356 participants; most participants were of minority ethnicities and females. Only one study noted if they conducted an a priori sample size. Gorlan and colleagues ²⁸ reported that with an effect size of .74, a power of 80%, and an alpha of .05, they would need at least 30 participants per condition. Six of the nine studies included an inclusion criterion of children at or above the age and gender-specific 85th percentile for BMI ^{24, 26, 27, 28, 29, 32}. All the retained studies were randomized control trials. The nine retained studies were varied in duration, location, and type of practitioner who conducted MI, and all but one study provided information regarding interventionist MI training ³². Interventionists included general members of the research team, ^{24, 26, 27, 30, 31} college-aged mentors ²⁵, a doctoral student ²⁸, and a health care provider ^{29, 32}. MI was the only framework for five out of nine studies ^{24, 26, 27, 29, 30}. Additional frameworks included social cognitive theory ^{25, 31}, self-determination theory, ²⁸ and transtheoretical model ³². The number of sessions ranged from 2 - 12 over 7 to 11 months. Seven out of nine studies implemented MI in person ^{24, 25, 27, 28, 29, 31, 32}, and two studies implemented MI via telephone calls. ^{26, 30} All the studies included a control or standard care group. Standard care centered around visits to health care providers and/or structured provision or access to health information. Adolescent participants were recruited from a variety of sources including schools ^{25, 27, 29, 31}, medical centers ^{24, 26, 28, 30}, digital media, television, and radio ³².

Five out of nine studies included PA as a primary outcome ^{24, 26, 28, 30, 31}. The other four studies had primary outcomes for changes in body composition ^{25, 27}, BMI ²⁹, and exercise capacity, defined as peak oxygen consumption ³². Of these four studies, all of them included PA as a secondary outcome. All PA outcomes are summarized below, followed by exercise capacity, body composition, and BMI outcomes, before other secondary outcomes.

Three studies found a statistically significant impact of the MI intervention compared to the control for PA outcomes ^{25, 28, 31}. Black and colleagues ²⁵ reported that overweight and obese participants in the intervention group saw a statistically significant increase in mean minutes of play-equivalent physical activity (PEPA) than overweight and obese participants in the control group over 11 months based on the ankle-worn accelerometer. PEPA was defined as 1,800 or more counts per minute ²⁵. These findings remained similar at a 24-month follow-up. Another study found that adding MI to a standard weight loss program (SWLP) across 6-months resulted in significant improvements in length of PA and energy expenditure, both measured subjectively and objectively than SWLP alone ²⁸. Finally, Neumark-Sztainer and colleagues ³¹ found that girls in the intervention had a statistically significant decrease in self-reported sedentary behavior across 25 weeks compared to girls in control. Additionally, girls in the intervention saw significant increases in stages of change for PA, PA goal setting, and self-efficacy to overcome PA barriers.

The remaining six studies did not find any statistically significant impacts of MI on changes in PA ^{24, 26, 27, 29, 30, 32}. Despite not finding differences in PA, Kong and colleagues ²⁹ found that participants significantly reduced TV viewing more than the control. While there were no differences between groups, two out of six studies found that the intervention and control groups significantly increased PA levels ^{24, 26}. Zanatta and colleagues ³² found that the intervention group significantly increased walking time but did not find a significant difference between groups.

While Zanatta and colleagues ³² did not find significant differences in PA outcomes, they did find that participants in the intervention had significant decreases in minute ventilation (V_E), which is defined as the amount of air breathed per minute, across three months compared to the participants in the control group. This finding indicates possible improvements in exercise capacity. However, other than V_E, no other significant differences between groups were found for any cardiopulmonary exercise test variables (i.e., peak oxygen consumption, oxygen consumption at anaerobic threshold, and ventilatory equivalents). Besides this variable, there were no other significant differences between groups for exercise capacity ³². McKillop and colleagues ³⁰ found no significant changes in physical fitness indicators in either the control or intervention group. These fitness indicators included anthropometric measures, aerobic fitness, flexibility, muscular strength, and muscular fitness.

Two authors found that MI had a statistically significant impact on body composition or BMI. Black and colleagues ²⁵ found that the intervention effectively reduced body fat percentage, fat mass and increased fat-free mass in overweight and obese youth. Another study found that eight MI sessions across an academic year resulted in a statistically significant reduction in BMI in the intervention group compared to the control ²⁹. The remaining seven studies did not find any significant impacts of MI on body composition or BMI when the intervention group was compared to the control ^{24, 26, 27, 28, 30, 31, 32}. Davis and colleagues ²⁷ found that circuit training (CT) alone successfully decreased waist circumference, visceral adipose tissue, and subcutaneous adipose tissue compared to the control. However, the addition of MI did not produce an additional benefit. Bean and colleagues ²⁴ found that MI and educational control groups saw significant reductions in BMI z-score, which remained significant after six months of no MI contact. Furthermore, while there were no between-group differences, Gourlan and colleagues ²⁸ did find that the MI intervention group significantly decreased BMI and perceived medical staff as more autonomy supportive.

Several studies focused on behavior change found that using MI significantly impacted dietary outcomes. Bean and colleagues ²⁴ found that the intervention group had a statistically significant reduction in caloric intake than the control group across three months. Additionally, Black and colleagues ²⁵ found that the intervention group had more significant decreases in daily consumption in most dietary categories than the control. Another study found that girls enrolled in an intervention utilizing MI techniques saw significant improvement in dietary goal setting, stage of change for fruit and vegetable intake, stage of change toward eating regular breakfast, increased portion control behaviors, and stage for portion control ³¹. Two authors did not find any group effects for dietary outcomes ^{27, 29}.

Other studies examined effects on clinical indicators. In addition to not finding any dietary behavior change outcomes, Kong et al. ²⁹ found that both the control and intervention groups saw increased fasting glucose. It was further noted that the control group had a significantly less increase compared to the intervention group. Besides this between-group difference, Kong and colleagues ²⁹ did not find that MI significantly impacted triglycerides, HDL-cholesterol, fasting insulin, HOMA-IR, and blood pressure. Davis and colleagues ²⁷ also found that MI did not significantly impact fasting glucose.

Two studies examined the impact on motivation or quality of life as secondary outcomes. Gourlan and colleagues ²⁸ did not find that adding MI to a weight loss program had an additional benefit for motivational regulation or perceived competence. When examining the impact of MI on an exercise intervention, McKillop and colleagues ³⁰ did not find an additional benefit for global quality of life or health-related quality of life.

All but one study ³² detailed how the interventionist was trained in MI; training time ranged from 7 hours to 2-weekend training courses. In three studies, authors specified that the interventionist was trained by an accredited organization, either the Motivational Interviewing Network of Training (MINT) ^{24, 30}, or the French Association of MI. ²⁸ In one study, the interventionists were trained members of the MINT ²⁷. Only three out of nine studies conducted a fidelity assessment. An additional study noted that they audio-recorded sessions and provided additional coaching four times throughout the intervention, but the authors did not assess fidelity ²⁹.

Of the three authors who did assess fidelity, they used the Motivational Integrity Treatment Integrity Scale. Bean and colleagues ²⁴ reported an intra-class correlation of 0.72 - 0.99. Davis and colleagues ²⁷ coded a subsample of 10% and had an average global rating of 3.8 and an average interclass correlation of 0.77. Finally, Gourlan and colleagues ²⁸ randomly selected 25 interviews to code and had an average global rating of 3.76. For reference, 3.5 is considered proficient, and 0.77 is in the excellent range. Despite not coding for fidelity, McKillop et al. ³⁰ and Neumark-Sztainer et al. ³¹ noted that supervised practice sessions and ongoing support in MI were delivered following their training.

The methodologic quality assessment ratings of the nine retained studies are presented in Table 2. All studies included a randomization process which reduces the risk of population bias. While the concealment procedure in one study ²⁷ was detailed, the other eight studies did not report this information, which may impact the risk of bias. Additionally, the MI-trained interventionist cannot be blinded to intervention conditions due to the nature of MI. Most of the studies included in this review were free of major selective reporting and other biases.

4. Discussion

This modified Cochran systematic review aimed to explore and report evidence and gaps in the literature regarding the impact MI has on PA levels in adolescents. The current literature suggests that MI is effective at improving health behaviors in adults, children, and adolescents. Martins and colleagues found that MI effectively increased diet and exercise self-efficacy and improved PA levels in adults ¹⁷. It is evident that similar benefits are found in children ¹⁹ and adolescents ¹⁸. The findings of our systematic review indicated that MI might be effective at improving PA levels in adolescents. Three out of nine retained studies found that MI had a statistically significant impact on PA and sedentary behaviors ^{25, 28, 31}. An additional study found that MI significantly impacted self-reported TV viewing ²⁹. Furthermore, Zanatta and colleagues ³² found that adding MI to a guided PA session successfully improved minute ventilation.

MI originated in addiction treatment but has since spread to examine health behaviors and target outcomes ^{17, 18, 19, 20}. One of these target outcomes includes weight management. In adults, MI is effective at improving BMI ³³. Similar results have been found for children and adolescents. A systematic review conducted by Suire and colleagues ²⁰ found that MI was effective at improving waist circumference, waist-to-hip ratio, and BMI in children under the age of 10. Furthermore, Crushing and colleagues ¹⁸ noted that MI interventions in adolescents successfully improve several health behaviors, including weight management. In the current systematic review, only two of the seven retained studies, which examined BMI and/or body composition, found MI successfully improved body composition or BMI in adolescents.

The current systematic review results indicate that MI might be an effective pathway to improving PA, BMI, and body composition. Additionally, it is essential to note that most of the included studies targeted adolescents at the highest risk (i.e., minorities, females, overweight, and obese) and were able to find significant improvements in PA in addition to BMI, body composition, and other weight management outcomes. Thus, MI might be a successful pathway to encourage PA, especially in populations at the highest risk for physical inactivity. However, due to the heterogeneity of the nine included studies, it is difficult to draw general conclusions regarding the impact MI has on PA behaviors in adolescents. Overall, this systematic review seems to support the current evidence that MI is effective at improving health behaviors in a wide range of age groups, including adolescents.

One of the most critical gaps in the literature is the variety of MI sessions, delivery, training, and fidelity. MI sessions ranged from 2 to 12, most of which were face-to-face; however, two studies utilized telephone calls as the delivery method ^{26, 30}. In the current systematic review, we found that the studies with a minimum of six 20-minute MI sessions found significant differences in PA regardless of delivery method ^{28, 31}. Thus, future studies should aim to have at least six 20-minute MI sessions to impact PA successfully. All but one of the studies included in this review detailed how the interventionists were trained. Interventionists received a minimum of 40 hours of MI training with ongoing support in seven of the eight studies that detailed interventionist training. The final study, which detailed interventionist training, noted that the research staff only completed seven hours of training in MI, which may be one explanation as to why this study did not find significant effects of MI on PA levels.

Furthermore, only three out of nine studies coded for fidelity, one of which found that MI had a significant effect on PA levels ²⁸. Fidelity assessments are crucial to understanding if what the participant or patient receives an evidence-based treatment. For researchers to understand the true impact MI has on PA levels, future studies should aim to be more consistent with training time and assessing fidelity to better understand the proficiency level needed to impact PA levels.

Although the methods employed in this study were valid and rigorous, the study is not without limitations. First, there was a high amount of heterogeneity in MI sessions, delivery, training, and fidelity. As a result, generalizations are limited. An additional limitation is heterogeneity in measuring PA. Some authors utilized self-report, some only did accelerometry or pedometers, and some did both.

Furthermore, with the self-reported survey, some authors utilized the 3-Day Physical Activity Recall (3-DPAR), one utilized the 7-Day Physical Activity Recall (7-DPAR), and one author utilized the International Physical Activity Questionnaire (IPAQ). Self-reported PA data increases the risk of participant bias ³⁴. In the current systematic review, we found that two out of the five studies which utilized self-reported surveys found significant between-group differences in PA ^{28, 31}. Whereas only one of the seven studies which utilized accelerometry, pedometers, or a combination of surveys and objective measures found significant between-group differences in PA ²⁵. Indicating that participant bias might have influenced PA outcomes. Future studies should aim to include an objective measure of PA.

A total of 392 articles were identified in our initial search. After removing duplicates, and screening, we were left with a total of nine studies included in this systematic review. While not without its limitations, this systematic review provides evidence that MI has the potential to improve PA behaviors in adolescents. Three authors found that MI was successful at improving PA behavior. Additionally, two authors found that MI was beneficial for body composition and BMI, which supports the current literature. To our knowledge, this is the first systematic review examining the impact MI has on adolescent PA levels. Results of this review indicate that MI is a possible pathway to improve PA levels, especially for at-risk youth. Future research would benefit from more standardized assessments for MI fidelity. Additionally, it would be beneficial to include objective measures of PA to decrease participant bias.

References