A Pilot Feasibility Study of Interpersonal Psychotherapy for the Prevention of Excess Weight Gain Among Adolescent Military-dependent Girls

ABSTRACT

Introduction

Adolescent military-dependents face unique psychosocial stressors due to their parents’ careers, suggesting they may be particularly vulnerable to excess weight gain and symptoms of depression and anxiety. Despite these risk factors, there is a lack of tested preventative interventions for these youths. Given the transient nature of military family deployments, research may be hindered due to difficulty in collecting long-term prospective outcome data, particularly measured height and weight. The primary aim of this study was to examine the feasibility and acceptability of collecting body mass index (BMI, kg/m²) outcome data up to 2 years following a randomized controlled pilot trial of an adapted interpersonal psychotherapy (IPT) program aimed at preventing excess weight gain and improving psychological functioning for adolescent military-dependents. In exploratory analyses, patterns in body composition over time were examined.

Materials and Methods

Twenty-seven adolescent military-dependent girls (baseline: M_age: 14.4 ± 1.6 years; M_BMI: 30.7 ± 4.9 kg/m²; M_BMI-z: 1.9 ± 0.4) participated in this study. After a baseline assessment, utilizing a computerized program to create a randomization string, girls were assigned to either an IPT or a health education (HE) program. Participants completed three follow-up visits (posttreatment, 1-year follow-up, and 2-year follow-up). Girls completed a Treatment Acceptability Questionnaire at posttreatment; at all time points, height and fasting weight were collected. For the primary aim, Fisher’s exact tests examined the rate of obtained follow-up data and lost to follow-up status between the two groups, Mann-Whitney U tests examined the session attendance between groups, and treatment acceptability ratings were compared between the two groups at posttreatment using an independent samples t-test. For the exploratory aim, one-way analyses of covariance (ANCOVAs) examined the group differences in BMI at each time point, adjusting for baseline values, and paired samples t-tests examined the within-group differences at each time point relative to baseline. Using imputed data in the full intent-to-treat sample, mixed model ANCOVAs were conducted to examine the group differences over time.

Results

Across both groups, girls attended an average of 72.0% of sessions. At least partial data were collected at posttreatment, 1-year follow-up, and 2-year follow-up for 96.3%, 85.2%, and 74.1% of the participants, respectively. There were no significant group differences in follow-up data collection rates, follow-up status, number of sessions attended, or treatment acceptability. BMI-z stabilized across groups, and there were no group differences in BMI-z. In adjusted ANCOVA models with imputed data, no significant group-by-time effects emerged.

Conclusions

For this randomized controlled prevention trial, long-term outcome data collection of measured BMI was possible in adolescent military-dependents and IPT was an acceptable and feasible intervention. An adequately powered trial is required to assess the efficacy of this intervention among military-dependents for obesity prevention and improvements in BMI.

INTRODUCTION

The rates of overweight (defined as a body mass index [BMI, kg/m²] ≥85th but <95th percentile adjusted for age and sex) and obesity (BMI ≥95th percentile) among adolescent military-dependents are nearing those of their civilian peers.¹ Yet, the adolescent children of military personnel may be particularly vulnerable to unique stressors that place them at higher risk for excess weight gain and emotional distress compared to civilian teens. Specifically, as a result of their parents’ careers, military youth contend with frequent relocations and parental deployments,² as well as the rigorous shape and weight standards experienced by their active duty caretakers.³ Concurrently, frequent relocations and cultural attitudes about mental and physical health may impede social support and help-seeking behaviors.⁴ Collectively, these unique stressors may increase susceptibility to mood-related psychopathology among military-dependents,^5–10 which may further contribute to and exacerbate excess weight gain.^11–14

Given that there are currently 1.68 million children with a parent in the military, 24% of whom are adolescents,¹⁵ and that the children of service members are themselves more likely to serve,¹⁶ attention to the health and psychosocial functioning of adolescent military-dependents is critical to both public health and military readiness. As overweight is the most common reason for rejection from the military,¹⁷ excess weight gain prevention among military-dependents may be essential in order to ensure the fitness of the future armed forces. However, there is currently a dearth of empirically tested interventions among military youth,¹⁸ particularly for the prevention of obesity. In part, the lack of research may be due to challenges in carrying out long-term studies and collecting distal outcome data in a transient population that is often lost to follow-up. Coupled with generally poor follow-up compliance in weight management programs,¹⁹ intervention studies among military populations are especially challenging.

Research in civilian populations suggests that interpersonal psychotherapy (IPT) may be effective in preventing excess weight gain in some adolescents at high risk for adult obesity.^20,21 Originally adapted from IPT for the prevention of depression²² and IPT for binge-eating disorder,²³ IPT for the prevention of excess weight gain posits that interpersonal difficulties, such as social skills deficits or role transitions common in adolescence, contribute to negative affect, which may result in non-homeostatic eating to cope with aversive emotional states. Recurrent overeating may then contribute to excess weight gain over time.²⁴ In a randomized controlled trial of civilian adolescent girls, IPT was not more efficacious than a standard-of-care health education (HE) group in decreasing expected BMI and fat gain at a 12-month follow-up.²⁰ However, among girls in high-risk groups, IPT was associated with significantly steeper decreases in BMI-z score, less adiposity gain,²¹ and improvements in non-homeostatic eating behaviors²⁵ by 3-year follow-up.^21,25 Considering the unique interpersonal stressors and high anxiety⁷ that adolescent military-dependents are vulnerable to, in conjunction with elevated risk for eating pathology,^5,9 IPT may be a particularly appropriate approach to examine in this population.²⁶

Therefore, we conducted a pilot feasibility trial among adolescent (12-17 years) military-dependent girls at risk for adult obesity. Military families move, on average, every 2 to 3 years, which corresponds to up to six relocations within a military child’s school career.² As such, a primary aim was to assess the feasibility of collecting follow-up study variables of interest for participants who had relocated over the course of the study period. Specifically, measured height and weight were collected via the military electronic health systems. Secondary outcomes were measures of feasibility and acceptability of the adapted IPT program. We hypothesized that participants in the IPT program would attend a greater percentage of sessions and rate the intervention as more acceptable than participants in a brief HE group. Lastly, we conducted exploratory analyses examining the relationship between assigned intervention condition (IPT vs. HE) and body composition.

MATERIALS AND METHODS

Participants

Female military-dependents (12-17 years at baseline) were recruited for a pilot study aimed at preventing obesity (ClinicalTrials.gov identifier: NCT02334202). Recruitment efforts were directed at parents through referrals from providers at Fort Belvoir Community Hospital; advertisements and flyers at local base facilities, clinics, websites, and listservs; in-person recruitment through information kiosks; and general mailings sent out to parents of adolescent girls who were enrolled in or eligible for military health insurance (Tricare) in the Greater Washington DC area.

Girls were eligible if they were at-risk for adult obesity due to a BMI ≥85th percentile for age and sex²⁷ and endorsed any disinhibited eating (loss-of-control over eating, eating in secret, emotional-eating, eating in the absence of hunger, eating in response to a negative trigger, feelings of guilt or shame after eating, or feelings of numbing out, zoning out, or spacing out while eating²⁸) during a telephone screening. Adolescents were excluded if they had a major, chronic medical or psychiatric illness other than binge-eating disorder; obesity-related medical complication; current or recent pregnancy; were taking medications that affect appetite, mood, or body weight; were involved in current psychotherapy or a structured weight loss program; or reported weight loss exceeding 3% during the 2 months before assessment. Girls taking oral contraceptives were included if the medication had been used for at least 2 months before starting the group program. The study was approved by the Uniformed Services University of the Health Sciences Institutional Review Board and the Fort Belvoir Community Hospital Department of Research Programs.

Procedures

After completing an initial phone screen, those who appeared eligible attended a baseline screening visit. Parents/guardians and girls provided written consent and assent, respectively. Following a baseline assessment, utilizing a computerized program to create a randomization string, girls were assigned to either an IPT or HE program. All in-person procedures and both groups were carried out in the Family Medicine Clinic at Fort Belvoir Community Hospital, where girls received their healthcare services.

As previously described,²⁰ IPT comprised 12-weekly 1.5-hour group meetings with two group leaders. The IPT program was adapted to address the unique challenges facing adolescent military-dependents, such as parental deployments and the stress of changing schools and friend groups due to relocations. The program included psychoeducation about risk factors for excessive weight gain and the provision of age-appropriate tools to address and resolve interpersonal difficulties and disputes, cope with negative mood states, and reduce eating in response to negative affect.

The brief HE program comprised four monthly 1.5-hour in-person group meetings with two leaders, and 12 weeks of material for participants to complete at home. The group utilized a workbook based upon the book 7 Habits of Highly Effective Teens²⁹ and provided information on how to live a healthier life (unrelated to eating). The workbook comprised modules on improving self-image, building friendships, resisting peer pressure, and achieving goals.

Leaders for both conditions were postdoctoral-level clinical psychologists, advanced graduate students in clinical psychology, or medical residents. Girls in both groups were asked to complete three follow-up visits after completion of the group program: (1) immediately after the conclusion of the group program (posttreatment), (2) 1 year after initiation of the group program (1-year follow-up), and (3) 2 years after initiation of the group program (2-year follow-up). At baseline and follow-up visits, height and fasting weight were measured. Follow-up visits were completed in-person, unless a participant had relocated during the study period. If so, girls were asked to visit a local Tricare clinic for measurement of height and weight. In such cases, a study team member accessed this height and weight from the Armed Forces Health Longitudinal Technology Application medical record system. Participants were compensated US$20 for each of the four assessment visits for a total possible amount of $80 for full participation.

Measures

Body Composition

Height and fasting weight were measured with calibrated instruments, once at each visit while the participant was lightly clothed, but not wearing shoes. BMI was computed, and BMI SD scores (BMI-z), accounting for age and sex, were calculated based on the CDC growth chart standards.²⁷

Treatment Acceptability Questionnaire

Intervention acceptability was assessed with the Treatment Acceptability Questionnaire³⁰ at the posttreatment assessment. This measure consists of six questions. Sample items include, “How effective do you think the groups were?”, “How trustworthy did you think the group leader was?”, and “Overall, how acceptable did you find the group meetings to be?” Participants rated each item on a scale of 1 to 7. Items were reverse-coded as necessary then items were summed such that higher scores demonstrate greater acceptability and tolerance of the intervention. The Treatment Acceptability Questionnaire is a reliable and valid measure for assessing acceptability of psychological interventions.³⁰ In the current sample, the questionnaire demonstrated fair internal consistency at posttreatment (Cronbach’s α = 0.76).

Data Analytic Plan

Analyses were conducted using SPSS 25.0 (IBM Corp, Armonk, NY) and SAS University Edition (SAS Institute Inc., Cary, NC). Data were screened for outliers and normality. To minimize outliers’ influence on findings, extreme outliers (defined a priori as >3 SD from the mean; <1% of data) were recoded to 3 SD from the mean.³¹ Baseline characteristics were compared across IPT and HE using independent samples t-tests or Chi-square analyses, as appropriate.

For the primary outcome, the rate of obtained follow-up data (presence vs. absence) for body composition at 1-year and 2-year follow-up was descriptively examined across the whole sample and compared between samples using a series of Fisher’s exact tests. In addition, a Fisher’s exact test was conducted to determine if there were differences in lost to follow-up status between IPT and HE at the 2-year follow-up visit. Given that the number of sessions differed in each condition, for the secondary outcome, the percentage of attended sessions were compared between participants in IPT and HE using a Mann-Whitney U test and a Fisher’s exact test was used to compare the proportion of participants in each condition who attended at least 75% (i.e., at least 3 of 4 sessions for the HE group and at least 9 of 12 sessions for the IPT group) of sessions. Further, treatment acceptability ratings were compared between the two groups using independent samples t-tests. For these analyses, missing data were handled using listwise deletion.

A one-way analysis of covariance was conducted to examine the group differences (IPT vs. HE) in BMI-z at posttreatment, 1-year follow-up, and 2-year follow-up, adjusting for the respective baseline value. Paired samples t-tests were then conducted within each intervention condition to examine within-group differences for these variables, relative to baseline. For these analyses, missing data were handled using listwise deletion. Additional analyses were conducted in the full intent-to-treat sample. Due to the small sample size and large amount of missing follow-up data, multiple imputation models failed to converge; therefore, missing data for these analyses were imputed using a last observation carried forward approach. Mixed model analyses of covariance were then conducted for the exploratory outcomes, using intervention condition and time as the independent variables and adjusting for age (years) and race (White = 1 versus non-White = 0). Due to violations of the sphericity assumption, the Greenhouse-Geisser correction was applied. Partial eta-squared, within-subjects Cohen’s d, and odds ratios (ORs) were calculated as the effect size for analyses, as appropriate. All tests were two-tailed and differences were considered significant when P values were <.05.

RESULTS

Thirty-eight girls were assessed for eligibility. Eleven were excluded: six did not qualify and five declined to participate. Therefore, twenty-seven girls were randomized. A consort diagram for participants is shown in Fig. 1. Thirteen girls were randomized to IPT while 14 girls were randomized to HE. There was no difference between the two groups in baseline age, race distribution, BMI, or BMI-z, ps >.15. Baseline sample characteristics are shown in Table I.

FIGURE 1.

Consort diagram for all study participants.

TABLE I.

Baseline Sample Characteristics

	Total sample (N = 27)	HE (n = 14)	IPT (n = 13)	P
Age (mean ± SD)	14.42 ± 1.56	14.56 ± 1.43	14.27 ± 1.73	.64
Race (%)				.59
Non-Hispanic White	25.90	35.70	15.40
Non-Hispanic Black	29.60	28.60	30.80
Hispanic	22.20	14.30	30.80
Multiracial	22.20	21.40	23.10
BMI-z (mean ± SD)	1.88 ± 0.41	1.77 ± 0.35	2.00 ± 0.46	.15

The P-value indicates group comparison without covariates (as assessed by independent samples t-tests or chi-square tests, as appropriate).

Abbreviations: BMI-z, body mass index adjusted for age and sex; HE, health education; IPT, interpersonal psychotherapy.

Data Collection through Follow-Up

Partial or full data were collected for 26 girls at posttreatment, 23 girls at 1-year follow-up (47.8% conducted remotely), and 20 girls at 2-year follow-up (70.0% conducted remotely). Fig. 1 illustrates consort diagram. BMI-z data were collected for 23 girls (85.2%) at 1-year follow-up and 19 (70.4%) girls at 2-year follow-up. When comparing the rate of obtained BMI-z data between IPT and HE, there was no significant difference at 1-year follow-up (IPT: 92.3%, HE: 78.6%; P = .60, OR = 3.27 [95% CI, 0.30-36.31]) or 2-year follow-up (IPT: 61.5%, HE: 78.6%; P = .42, OR = 0.44 [95% CI, 0.08-2.38]).

When comparing loss to follow-up status at 2-year follow-up, there was also no difference between the two groups, P = .68, OR = 0.61 (95% CI, 0.11-3.49). When lost to follow-up status was examined across both groups using logistic regressions, baseline age was inversely associated with lost to follow-up status (P = .02, OR = 0.40 [95% CI, 0.18-0.87]). However, neither baseline BMI-z (P = .51, OR = 2.11 [95% CI, 0.23-19.45]) nor race (P = .45, OR = 2.05 [95% CI, 0.32-13.16]) were significantly associated with lost to follow-up status.

Acceptability and Feasibility of IPT

The average percentage of sessions attended for all participants was 72.0%. For IPT, the median number of sessions attended was 9 (out of 12; 75.0%), while for HE, the median number of sessions attended was 3 (out of 4; 75.0%). There was no significant difference in the percentage of sessions attended between the two groups, U = 73.00, P = .40. When comparing the number of participants who attended at least 75% of the sessions, there was no difference between IPT (61.4%) and HE (71.4%; P = .70, OR = 0.64 [95% CI, 0.13-3.20]). For treatment acceptability ratings, there was no difference between IPT (M = 39.45, SD = 2.02) and HE (M = 37.25, SD = 4.79; t(15.05) = 1.46, P = .17, d = 0.60).

BMIz

Adjusting for baseline BMI-z, there were no significant differences in BMI-z between IPT and HE at posttreatment (IPT: M_adj = 1.91, SE = 0.03; HE: M_adj = 1.88, SE = 0.03; P = .48, η_p² = 0.02), 1-year follow-up (IPT: M_adj = 1.98, SE = 0.05; HE: M_adj = 1.96, SE = 0.05; P = .75, η_p² = 0.01), or 2-year follow-up (IPT: M_adj = 1.81, SE = 0.07; HE: M_adj = 1.96, SE = 0.05; P = .15, η_p² = 0.13). Within HE, BMI-z was not significantly different from baseline (M = 1.77, SD = 0.35) at posttreatment (P = .92, d = 0.03), 1-year follow-up (P = .84, d = 0.06), or 2-year follow-up (P = .86, d = 0.06). For IPT, BMI-z was also not significantly different from baseline at posttreatment (P = .07, d = 0.56), 1-year follow-up (P = .54, d = 0.19), or 2-year follow-up (P = .15, d = 0.57). Within-group data at each time point are shown in Table II. Examining the imputed data, there was a significant main effect of time [F(2.12, 48.82) = 4.92, P = .01, η_p² = 0.18], but no significant main effect of group [F(1, 23) = 1.26, P = .27, η_p² = 0.05], or group-by-time interaction [F(2.12, 48.82) = 0.26, P = .79, η_p² = 0.01] for BMI-z.

TABLE II.

Exploratory Analyses: Within-Group Paired Samples t-Tests Relative to Baseline

	HE Change (Mean ± SD)	df	t	P	d	IPT Change (Mean ± SD)	df	t	P	d
BMI-z
Post-Treatment	−0.003 ± 0.11	12	−0.11	.92	0.03	0.03 ± 0.06	12	1.99	.07	0.56
1-Year	0.01 ± 0.16	10	0.20	.84	0.06	0.03 ± 0.15	11	0.64	.54	0.19
2-Year	−0.01 ± 0.13	10	−0.19	.86	0.06	−0.11 ± 0.19	7	−1.62	.15	0.57

Statistical data from paired samples t-tests.

Abbreviations: BMI-z, body mass index adjusted for age and sex; HE, health education; IPT, interpersonal psychotherapy.

DISCUSSION

In this pilot feasibility study of adolescent military-dependent girls at-risk for adult obesity, retention and data collection of measured height and weight throughout the follow-up period was high. This is notable considering the itinerant nature of the military family, and difficulties in obtaining long-term objective BMI data in weight management trials,¹⁹ indicating that the format of the study and flexibility of data collection methods for height and weight were successful.

The pattern of findings suggests that BMIz stabilization in IPT did not differ from HE. However, across groups, most girls experienced a stabilization in BMI-z. Given that youth with above average weight are at high risk for gaining excess weight as they grow,^32,33 it is possible that both interventions could have had a positive impact. However, without a contemporaneous control group, this possibility must be interpreted with caution. These results are consistent with previous research in civilian samples, suggesting that both IPT and HE may provide social and community support sufficient to reduce gains in BMI-z.²⁰ The group cohesion experienced in either group may be especially beneficial to the children of military service members, who frequently relocate due to their parents’ careers. Furthermore, when considering the itinerant nature of the military family, as well as the common experience of weight-based teasing among adolescent dependents with overweight,³⁴ these youths may lack the continuity and support in their community, school, and friend groups afforded to civilian children. Therefore, the group cohesion and communality resulting from both groups may have been beneficial to all participants. However, a fully powered trial should be conducted to examine the impact of IPT on BMI-z over time.

A key study strength includes the flexible follow-up procedures for height and weight data collection, which enabled BMI determination through annual doctor’s visit or scheduled appointment, even if participants had relocated over the course of the study period. Such approaches might be considered for other weight management trials in military populations, which typically suffer from poor retention and follow-up,³⁵ similarly to trials in civilian populations.¹⁹ Other strengths include the assessment of a vulnerable, high-risk group that is frequently neglected in intervention research. Further, the sample was diverse, with a majority identifying as a racial or ethnic minority. In accordance with best practices,^36,37 both completer and intent-to-treat analyses were conducted for all study outcomes. The study also utilized objectively measured height and weight, although it is possible that measurements retrieved from the electronic medical records were not collected with the same precision of those collected by the study team.

A primary study limitation is that the two groups were not matched for time and attention; the IPT group met weekly for 12 weeks, whereas HE met monthly, for a total of four in-person sessions. It is possible that any patterns suggesting enhanced effectiveness of IPT over HE in some domains were an artifact of the additional time and attention afforded to girls in this condition. Furthermore, baseline age was inversely associated with lost to follow-up status, suggesting that the flexible follow-up procedures were not as effective for older youth that may have aged out of the military system.

Child military-dependents are more likely to serve in the military than their civilian counterparts¹⁶; over 75% of new military recruits have at least one family member who served.³⁸ Thus, maintaining the health and fitness of the military family is imperative for both public health and military readiness. As these youths appear to be at high-risk for adverse psychological and health outcomes, potentially due to unique psychosocial stressors as a result of their parents’ careers,³ testing interventions to improve weight outcomes are warranted. Future research is required to determine whether IPT is effective in reducing obesity among this vulnerable and understudied population.

FUNDING

USU 72NC-01 to T. Sbrocco, NIDDK 1R01DK104115-01 to M. Tanofsky-Kraff, the Defense Health Agency, MED 83-10180 to M. Haigney. J. A. Yanovski is supported by the Intramural Research Program, Eunice Kennedy Shriver National Institute of Child Health and Human Development, ZIA-HD-00641. The funding sources had no involvement in study design, in the collection, analysis and interpretation of data, in the writing of the report, or in the decision to submit the article for publication.

CONFLICT OF INTEREST STATEMENT

None declared.

DECLARATIONS OF INTEREST

None declared.