Reciprocal Prospective Relationships Between Loneliness and Weight Status in Late Childhood and Early Adolescence

Reciprocal Prospective Relationships Between Loneliness and Weight Status in Late Childhood and... Adolescents who do not conform to weight ideals are vulnerable to disapproval and victimization from peers in school. But, missing from the literature is a prospective examination of weight status and feelings of loneliness that might come from those experiences. Using data from the Québec Longitudinal Study of Child Development, we filled that gap by examining the prospective associations between loneliness and weight status when the sample was aged 10–13 years. At ages 10, 12, and 13 years, 1042 youth (572 females; 92% from French speaking homes) reported on their loneliness and were weighed and measured. Family income sufficiency was included in our analyses given its relationship with weight status, but also its possible link with loneliness during early adolescence. The findings showed that (1) weight status and loneliness were not associated concurrently; (2) weight status predicted increases in loneliness from ages 12 to 13 years; and (3) loneliness predicted increases in weight from ages 12 to 13 years among female adolescents, but weight loss among male adolescents. The fact that loneliness was involved in weight gain for females suggests that interventions focused on reducing loneliness and increasing connection with peers during early adolescence could help in reducing obesity. ● ● ● ● ● Keywords Loneliness Body mass index Coping Income sufficiency Socioeconomic status Obesity Introduction et al. 2016), heightened loneliness among 10–14 years olds (Hayden-Wade et al. 2005) and weight concerns in pre- Social relationships and weight are prevalent concerns adolescents (Sinton et al. 2012). However, the prospective during adolescence (Denneel et al. 2018; Markey 2010). relationship between obesity and loneliness has yet to be Adolescents are vulnerable to peer disapproval of body size examined. (Lawler and Nixon 2011), feelings of loneliness (Qualter Building on longitudinal work that not only indicates et al. 2015) and they are driven by a need to “fitin” (Reitz that obesity predicts depression, but that depression also et al. 2014). Given the stigma associated with being predicts obesity (Goldschmidt et al. 2010), we hypo- overweight or obese, peer disapproval is high, creating thesized a bidirectional relationship between loneliness and negative social consequences (Harrist et al. 2016; Puhl high weight status. Conversely, feelings of loneliness are reported by those with eating disorders characterized by low body weight (Puhl and Suh 2015) and a meta-analysis indicated that low weight and depressive symptoms are Electronic supplementary material The online version of this article (https://doi.org/10.1007/s10964-018-0867-9) contains supplementary bidirectionally-related over time (Puccio et al. 2016). material, which is available to authorized users. Therefore, the current work with adolescents pro- spectively examines the relationship between (a) loneliness * Pamela Qualter and high weight status and (b) loneliness and low body pamela.qualter@manchester.ac.uk weight. Manchester Institute of Education, University of Manchester, Oxford Road, Manchester M13 9PL, UK The Social Context of Weight Status The School of Psychology, The University of Central Lancashire, Preston, UK The bio-ecological framework highlights two contexts Université Laval, Québec, QC G1V 0A6, Canada important for adolescent health because they influence The University of Montreal, Québec, Canada beliefs and behavior: (1) the immediate peer group, and (2) 1234567890();,: 1234567890();,: 1386 Journal of Youth and Adolescence (2018) 47:1385–1397 the wider social context within which the individual and (Rankin et al. 2016; Puccio et al. 2016). Findings show that peers live (Bronfenbrenner 2005). Social exclusion by peers obese 12–14 year olds have increased chance of developing in school and the accompanying feelings of loneliness depression (Eschenbeck et al. 2009), but also that depres- during adolescence are recognized as significant influences sive symptoms predict weight gain (Goldschmidt et al. of adolescent health (Hawkley and Capitanio 2015). But, 2010). The same bi-directional effects have been found for social exclusion based on weight status needs to also be those with low weight status too, with low weight predict- understood within the wider social context, where social ing depressive symptoms, but also depressive symptoms norms of the ideal body size create stigma associated with predicting decreases in weight (Puccio et al. 2016). The non-ideal weight, influencing adolescents to criticize their effects appear to be more pronounced for adolescent girls peers’ appearance (Lawler and Nixon 2011) and tease them compared to boys (Anderson et al. 2006). for non-conformation to weight ideals (Mooney et al. 2009). Mechanisms linking depression with obesity (Gold- While male and female adolescents are often criticized schmidt et al. 2017) and low weight (Puccio et al. 2016) about their appearance by their peers to a similar degree during adolescents have been examined. That evidence (Lawler and Nixon 2011), the social norms of ideal weight shows that distress caused from the feelings of shame and are different for males and females because gender stereo- guilt of not conforming to the body and weight ideal con- types of the socio-cultural ideal of beauty emphasize thin- tribute to obesity and low weight being linked over time ness for women and female adolescents (Puhl and Brownell with depression. Goldschmidt et al. also showed that 2001). Such findings suggest that female adolescents may depressive symptoms predict engagement in emotional be more vulnerable to the social context of appearance than eating as a way to alleviate distress. male adolescents, receiving greater pressure from peers to conform to the socio-cultural ideals. Empirical evidence Weight Status and Loneliness shows that females are particularly vulnerable to the nega- tive social effects of high weight status, experiencing more In contrast to the work exploring weight status and rejection and victimization from the peer group than boys depressive symptoms during adolescence, there is a paucity after the age of seven years (Qualter et al. 2015). Thus, of research examining concurrent and prospective rela- gender-based social norms are important to consider when tionships between weight status and loneliness at that same examining the interaction between weight status and lone- development period. Loneliness is the negative feeling that liness because females may be particularly vulnerable to the occurs when a person does not perceive their social rela- negative social effects of not conforming to weight status tionships to be as satisfying as they would like (Perlman ideals, experiencing, as a result, more loneliness than boys. and Peplau 1981), sharing one common symptom with Other society level influences are also important to depression (negative affect). But, loneliness is determined, consider in the current work. Socioeconomic adversity specifically, by a negative emotional response to a lack of during childhood and adolescence is linked to develop- close affiliative ties to peers during adolescence; depres- mental processes: low income is a known risk factor for sion, in contrast, is attributed to a broader range of obesity earlier in development (Grow et al. 2010) because it causes, including determinants other than deficient social leads to stress responses that exacerbate metabolic pro- relations (i.e., the individual’s genetic makeup, neurological cesses, leading to increased weight status (Wickrama et al. disorders, psychological dysfunctions; Koenig and Abrams 2013). Further, there is reason to think that low income may 1999). Thus, an exploration of the prospective association also be linked to loneliness. To our knowledge, there is between weight status and loneliness is important given currently no research on socioeconomic status and lone- that loneliness specifically taps distress associated with liness during adolescence, but parents who have limited peer group problems, which are hypothesized to be a sig- resources and income may not find adequate time to spend nificant part of the puzzle linking weight status with with their children, contributing to increasing distance depression. between parents and their children and child negligence; Research exploring the mechanisms linking weight status they also may not have the financial resources to ensure and depression discusses (1) how peer problems are a cat- their child’s engagement in specific peer group activities alyst for depressive symptoms among adolescents with the that cost money. For those reasons, we include family highest and lowest weight statuses, but also (2) how coping income sufficiency in our analyses. with emotional distress can lead to emotional eating and increases in weight status. Based on the fact that loneliness Weight Status and Depressive Symptoms includes the same negative affect as depression, we hypo- thesize a prospective bidirectional association between Being at the extremes of weight status (overweight/obese or loneliness and weight status, which we discuss further underweight) is associated with depressive symptoms below. Journal of Youth and Adolescence (2018) 47:1385–1397 1387 Weight Status Predicting Loneliness empirical studies. Consistent with the affect-regulation model of binge-eating (Heatherton and Baumeister 1991), It is clear, from empirical research, that there is stigma attached evidence shows that induced loneliness leads to increased to high weight status. Empirical evidence shows that, in eating as a way to alleviate distress (Rotenberg and Flood western society, the high degree of social stigma attached to 1999). But, loneliness has also been shown to increase obesity/overweight is evident as early as pre-school (Turnbull stress hormones such as cortisol that come from heightened et al. 2000), with increasing negative ratings for “chubby” physiological stress during episodes of loneliness body types by children aged 5–10 years (Brylinsky and Moore (Cacioppo and Hawkley 2003), and those stress hormones 1994) and peer social rejection of obese children at ages 6–7 affect fat storage and transportation with the body (Dallman years(Harristetal. 2016). Thosenegativesocial circumstances 2010; Moyer et al. 1994). continue into adolescence, with obese and overweight ado- In addition to the direct weight-related effects of stress lescents being regularly stigmatized, socially excluded, and hormones, a combination of perceived stress, disturbed victimized by peers (Puhl and Luedicke 2012;Puhletal. sleep, and cognitive rumination resulting from loneliness 2016). Thus, high weight status is associated with poorer peer (Zawadzki et al. 2013) could affect eating behavior by relationships, with overweight 10–14 year olds reporting impairing one’s ability to address problems, leading to higher levels of loneliness (Hayden-Wade et al. 2005). Given greater use of passive coping strategies, such as emotional that evidence, one could posit that higher weight status would eating (Nolen-Hoeksema et al. 2008). Those loneliness- result in less satisfaction with social relationships over time related stressors are also argued to increase our seeking of i.e., greater feelings of perceived loneliness. To date, however, easy learned-rewards such as high fat and high sugar foods the prospective link between higher weight status and feelings due to stimulation of the motivation and reward circuits in of loneliness has not been examined. the brain (Dallman 2010; Hanlon et al. 2016), making it In addition to predicting that higher weight status could easier to gain weight and harder to lose it. result in greater loneliness, it is also possible that those with The work represented above supports the idea that low weight status will report increasing loneliness over loneliness may lead to increases in weight status, but the time. Empirical work by Wang et al. (2010) shows that male prospective link has not yet been examined. The current and female adolescents with high weight status are often study examines whether loneliness predicts increases in bullied by their peers, but underweight girls are also often weight during late childhood/early adolescence. victims. That aggression, and the accompanying feelings of shame and social isolation from peers, is thought to be a consequence of intrasexual competition that is promoted by The Current Study society’s emphasis on thinness among females (Vaillan- court 2013). Rotenberg et al. (2013) and Rotenberg and An examination of the reciprocal prospective relationships Sangha (2015) have also shown that adolescents with low between loneliness and weight status among adolescence is weight status as a result of eating pathologies report higher important if we want to offer effective intervention solu- levels of loneliness compared to their peers, a consequence tions for loneliness and obesity among youth. In the current of stigmatization and alienated from the peer group. study, feelings of loneliness and weight status are explored Given the work detailed above, there is a need to in a population sample of Canadian 10–13 year olds, and examine whether both high and low weight status predict we examine whether high and low weight status put ado- increasing loneliness over the adolescent years. We pos- lescents at risk of later loneliness, but also whether lone- tulate that there will be concurrent and prospective asso- liness scores predict increasing weight status. The findings ciations between weight status and loneliness, and that could have important implications for interventions for those relationships will be curvilinear in nature, such that obese or underweight youth, and for adolescents who report individuals with either the highest or lowest weight sta- loneliness. Guided by the bioecological framework, we also tuses will feel more loneliness during adolescence com- seek to examine gender and income sufficiency as mod- pared to their normal weight peers. The current study will erators of the prospective relationships between weight examine the prospective curvilinear associations between status and loneliness. weight status and loneliness. Loneliness as a Predictor of Increasing Weight Method Status This study utilized data from the Québec Longitudinal Among adults, loneliness has been tentatively implicated as Study of Child Development (QLSCD), a large on-going a risk factor for increased weight, but there are few study which has tracked the health and wellbeing of a 1388 Journal of Youth and Adolescence (2018) 47:1385–1397 Table 1 Characteristics of the Time point Age of participants in months n = Sample at the Start of the (SD) Quebec Longitudinal Study of Child Development (QLSCD) Start of QLSCD survey in 1998 4.5 (0.55) 2120 survey and at T1–T3 of the T1 current study 121.70 (3.10) 1042 current study T2 current study 145.60 (3.05) 1042 T3 current study 157.60 (3.12) 1042 Demographics taken at start of QLSCD for the current sample compared to original sample Sample chosen for QLSCD Sample in current study Males (%) 51 45 French-speaking families (%) 81 92 Mother’s age in years 29 31 Father’s age in years 31 33 Mothers did not hold a high school degree (%) 17 15 Fathers did not hold a high school degree (%) 20 16 Mother had a university degree (%) 28 32 Father had a university degree (%) 26 27 Income Sufficiency at normal levels 81 88 Households headed by single parent (%) 7 5 Statistics Canada’s low-income cut-offs (LICOs) were used to categorize the families of participants on income sufficiency. LICOs are income thresholds at which a family would typically spend 20% more of its income than the average family on the necessities of food, shelter, and clothing. Families are classified as having “sufficient income” when the household income is above the low-income threshold determined by Statistics Canada in any given year. When income is between 60 and 90% of the low-income threshold, households are classified as having “insufficient income”; income levels below 60% of the low-income threshold are considered as “very insufficient” QLSCD Quebec longitudinal study of child development random sample of Quebec infants on a range of measures T1–T3 of the current study. Participants with and without since they were 5 months old (see the study website all data for the period of the current study were compared http://www.iamillbe.stat.gouv.qc.ca/default_an.htm for fur- using Little’s missing completely at random test (Little ther details). The representative sample is made up of 1988) to determine whether data imputation would be children born throughout 1998 in the Canadian province of possible. That comparison resulted in a significant chi- Québec (total population over 7 million, with approximately square value, Χ (18) = 35.246, p = .009, suggesting that 70,000 newborns per year). A total of 2,940 infants were missing values could not be dealt with using data imputa- selected for QLSCD through a region-based stratified tion methods. Thus, we used listwise deletion of cases, sampling design, of which 2120 infants (48.8% girls) took analyzing data from 1042 participants (572 females, 470 part, with parents providing informed consent in 1998. males) for whom complete loneliness, height, and weight Twins and children with major diseases at birth were not data were available at all three time points . Table 1 shows part of the study. In the current study, four waves of data how the final sample in the current study compares to the were used from successive data collections when the sample original QLSCD sample that had been chosen as a repre- of children was aged 10, 12, and 13 years. We refer to these sentative sample of children in Québec in 1998. The table time points as T1, T2, and T3 respectively of the current shows that, while there were fewer males in the current study. The QLSCD was approved by the Health Research Ethics Committees of the Québec Statistics Institute and the Loneliness, weight, and height data were also available for 926 of University of Montreal. the 1042 children at a further time point when the sample was aged 15 years. However, the measure of loneliness was limited to a single item, with response items that did not align with that for the earlier lone- Participants liness items; height and weight data were self-reported. In addition, there were so few participants who were classified as underweight (N Loneliness, weight, and height data were collected from = 20 females; 9 males) or obese (N = 9 females; 5 males) that it would 1259 children/early adolescents (667 females, 592 males) at not make the analyses viable. Journal of Youth and Adolescence (2018) 47:1385–1397 1389 sample than in the original sample, the reduced sample recommendations from The International Obesity Task taking part in the current study was representative of the Force BMI (IOTF; Cole et al. 2007). Those recommenda- same children living in the province of Québec in 1998. tions classified participants into the following weight cate- gories at each time point: thin grade 3 (BMI ≤ 16), thin Measures grade 2 (BMI ≤ 17 & > 16), thin grade 1 (BMI ≤ 18.5 & > 17), overweight (BMI ≥ 25 & < 30), obese (BMI ≥ 30) Loneliness morbid obesity (BMI > 30), and normal weight (BMI ≥ 18.6 & ≤ 24.5). Those IOTF cut-offs are recommended for Loneliness was measured using the Loneliness and Social international research and comparison (Lobstein et al. Satisfaction questionnaire developed by Rotenberg et al. 2004), and are used here to create weight status groups for (2004). This three-item measure is similar to the 3-item each time point, enabling us to explore whether different short form of the UCLA (University of California, Los weight categories reported higher levels of loneliness at Angeles) Loneliness Scale (Hughes et al. 2004), but the each time point. word “isolated” was simplified to “alone” in item 3. Items asked the extent to which, in the last two weeks, participants Income sufficiency had felt (1) “they had people they could talk to”, (2) “left out of things” and, (3) “alone”. Item 1 was reverse coded so Sufficiency of income was determined by low-income cut- that higher scores represented higher feeling of loneliness. offs set by Statistics Canada in any given year. It takes into Participants answered how they best described those feel- account household income, the size of the household, and ings (1 = never, 2 = sometimes, 3 = always), with possible the size of the residence area. Statistics Canada’s low- total scores ranging from 3 to 9. Cronbach’s alpha for the income cut-offs (LICOs) are income thresholds at which a loneliness measure was .66, .68, and .74 at T1, T2, and T3 family would typically spend 20% more of its income than respectively. Total scores on the loneliness scale were used the average family on the necessities of food, shelter, and in our regression analyses looking at the prospective asso- clothing (https://www.statcan.gc.ca/pub/75f0002m/2012002/ ciations between loneliness and weight status, but we also lico-sfr-eng.htm). Families are classified as having “sufficient created loneliness groups to examine whether children who income” when the household income is above the low- reported higher levels of loneliness at any given time also income threshold determined by Statistics Canada. When had higher weight status scores (higher BMI) at that same income is between 60 and 90% of the low-income thresh- time point. Following Yang and Victor (2011) we created a old, households are classified as having “insufficient “frequently lonely” group of participants who scored 6 or income”; income levels below 60% of the low-income above on the loneliness scale; all remaining youth were threshold are considered as “very insufficient”. Although grouped into a “not lonely/occasionally lonely” category. LICOs are widely used, they do not measure poverty. Those groups were created for the purpose of making Unlike the US, Canada does not have a measure of poverty. comparisons on BMI at each time point. For example, according to Statistics Canada (https://www. statcan.gc.ca/pub/75f0002m/75f0002m2010003-eng.htm) Weight status low-income thresholds are different for a family living in a rural area compared to similar families living in large cities. Trained research assistants weighed and measured the par- ticipants when they were wearing lightweight clothing and Analyses Plan no shoes. Two measurements were taken, and, if they varied by more than 0.5 cm for height or 0.2 kg for weight, a third First, using ANOVA, we examined whether there were measurement was taken. Where multiple measures were differences between people in the different weight cate- available, we used the average of each measure to calculate gories on loneliness at each time point. Second, using T- BMI (BMI = kg/m2). The computation of the participants’ tests, we examined differences in BMI between groups of BMI was followed by the creation of a BMI z-score using youth identified as either “frequently lonely” or “not lonely/ respondents’ BMI, self-reported age, gender, and the occasionally lonely”. We conducted all analyses separately external reference sample from WHO (Cole et al. 2000;de by gender. Third, we conducted a series of chi-square tests Onis et al. 2007). Those BMI z-scores (referred to here as z- to clarify the associations between loneliness and weight BMI) were used in our regression analyses, exploring the status, exploring for males and females separately, mem- prospective association between weight status (z-BMI) and bership of any given loneliness group by weight category at loneliness. As well as create a z-BMI score for each parti- each time point. Those chi-square analyses helped us to cipant, we also classified each of them as underweight, understand whether adolescents who were “frequently overweight, obese, or normal weight, following lonely” at any time point were more likely than chance to 1390 Journal of Youth and Adolescence (2018) 47:1385–1397 Table 2 Mean (and standard deviations) for loneliness in each weight category at each time point for females and males Time (age) Underweight Normal range Overweight Obese BMI < 18.5 18.51–24.99 BMI ≥ 25 BMI ≥ 30 Females Males Females Males Females Males Females Males T1 (10 year) 3.64 (.93) 43.46 (.95) 3.82 (1.19) 3.88 (1.23) 3.97 (1.33) 3.86 (1.29) 4.03 (1.27) 3.72 (.96) N = 1042 36 26 407 325 99 90 30 29 T2 (12 year) 3.71 (.86) 3.28 (.61) 3.76 (1.19) 3.81 (1.25) 3.90 (1.17) 3.73 (1.09) 3.97 (1.17) 3.58 (.97) N = 1042 35 25 384 311 116 98 37 36 T3 (13 year) 3.80 (1.10) 3.93 (1.51) 3.83 (1.31) 3.92 (1.29) 3.85 (1.14) 4.06 (1.43) 3.84 (1.26) 3.82 (1.29) N = 1042 30 28 384 299 120 105 38 38 Notes: Sample participants were categorized using the following recommendations from The International Obesity Task Force BMI (IOTF; Cole et al. 2007): thin grade III (BMI ≤ 16), thin grade II (BMI ≤ 17 & > 16), thin grade I (BMI ≤ 18.5 & > 17), overweight (BMI ≥ 25 & < 30), obese (BMI ≥ 30), morbid obesity (BMI > 30), and normal weight (BMI ≥ 18.6 & < 24.5). Underweight = Grade I, II, III were combined due to low numbers in Grades II and III; Obese (obese and morbid obese were combined due to low numbers in the morbid obese category); Possible loneliness scores ranged from 3 to 9. N = 1042 (Female = 572; Male = 470). ANOVAs revealed no differences between males and females on feelings of loneliness at each time point also be members of the obese or underweight weight F (3, 466) = 1.835, p = .140; and T3 (age 13 years) groups. Together, those sets of analyses enabled us to females, F (3, 568) = .014, p = .998, and males, F (3, 466) examine concurrent relationships between loneliness and = .393, p = .758. Thus, in the current population sample, it weight status. seems weight status is not associated with concurrent Fourth, we examined prospective linear and curvilinear reports of higher loneliness during adolescence for males or relationships between (1) z-BMI and loneliness, and (2) females; those with high or low weight status did not report loneliness and z-BMI using Hierarchical Regression Ana- higher levels of loneliness compared to normal weight lyses (HRAs), controlling for gender and income suffi- category to report feelings of loneliness (Table 2). ciency. Curvilinear analyses were used to determine the The next set of analyses examined whether adolescents exact associations between weight status at one time classified as “frequently lonely” were different to peers point and loneliness at the following data collection categorized as “not lonely/sometimes lonely” on weight waves, and loneliness at one time point and weight status at status at each of the time points. We ran a series of inde- the other data collection waves. Based on our findings pendent t-tests, separately for males and females and found of quadratic effects between z-BMI and loneliness, no difference between the “frequently lonely” and “not Structural Equation Modelling was not an appropriate lonely/sometimes lonely” groups at age 10 years (T1: analytic tool and the HRAs are presented as our final sta- females, t (74.78) = 1.88, p = .064; males, t (465) = 1.12, p tistical analyses. = .264) or age 12 years (T2: females, t (570) = 1.26, p = .210; males, t (467) = .03, p = .974). At age 13 years (T3), there was a difference between the lonely groups for Results males (t (81.38) = 3.04, p = .003), but not females (t (77.77) = 1.53, p = .131). Findings showed that male ado- We examined differences in mean loneliness for young lescents aged 13 years in the “frequently lonely” group had people in different weight categories using ANOVAs, a higher weight status than their male peers in the “not looking at each time point and males and females sepa- lonely/sometimes lonely” group (Table 3). rately. Because there were so few participants in the three Given that the mean BMI scores at T3 for males in the grades of thinness as defined by (Cole et al. 2007), we “frequently lonely” group would be considered within the merged those groups to create one group that we defined as normal range of BMI scores according to The International “underweight” (BMI < 18.5). We also merged the obese and Obesity Task Force BMI cut-offs (Cole et al. 2007; normal morbidly obese groups given that there were so few parti- weight = BMI ≥ 18.6 & ≤ 24.5), we decided to explore the cipants classified as morbidly obese. concurrent association between loneliness and weight status The one-way ANOVAs investigating loneliness by at T3 further. We conducted two Fisher-Freeman-Halton weight category revealed no significant differences at any of tests, one for males and one for females. Those analyses the three time periods: T1 (age 10 years) females, F (3, 568) showed that, at age 13 years (T3), males classified as = .994, p = .395, males, F (3, 466) = 1.057, p = .367; T2 “frequently lonely” were more likely than chance to be in (age 12 years) females, F (3, 568) = .741, p = .528, males, the normal weight category, while boys in the “not lonely/ Journal of Youth and Adolescence (2018) 47:1385–1397 1391 Table 3 Mean (and standard deviations) for BMI by loneliness group A further set of HRAs was conducted to examine the at each time point for females and males longitudinal over-time effects of loneliness on BMI. The Time Frequently lonely Not lonely/Sometimes lonely regressions followed the pattern and procedure outlined Females Males Females Males above, with z-BMI at each time point as the independent variable, and loneliness and z-BMI from earlier time points T1 (10 year) 19.26 (4.17) 17.76 (2.32) 18.28 (3.11) 18.25 (2.06) as predictors. Any two-way interactions between z-BMI x N = 1042 66 54 506 416 T2 (12 year) 20.63 (3.43) 19.86 (3.92) 19.98 (3.88) 19.84 (3.76) Gender were further examined by testing for the linear and N = 1042 60 49 512 421 quadratic (curvilinear) relations on the measure for each T3 (13 year) 21.82 (5.39) 22.52 (5.11) 20.80 (3.84) 20.56 (3.79) gender separately. Key information from the HRAs are N = 1042 69 69 503 401 detailed in the manuscript text, with tables (Tables S1-S6) detailing all HRA information included as on-line supple- Notes: Those in the “Frequently Lonely” group scored 6 or above on the loneliness scale; those scoring between 3 and 5 were classified as mentary information. “Not Lonely/Sometimes Lonely”;N = 1042 (Female = 572; Male = 470); T-tests showed no differences on BMI between same sex peers Stability of Weight and Loneliness Over Time in the “Frequently Lonely” and “Not Lonely/Sometimes Lonely” groups at ages 10 and 12 years. At age 13 years, males in the “Frequently Lonely” group scored significantly higher on BMI As anticipated, over time the strongest predictor of z-BMI compared to their same sex peers in the “Not Lonely/Sometimes and loneliness were previous measures of the same con- Lonely” group; there were no significant differences between females. struct, confirming stability. The strongest effects for both All BMI mean scores would be considered in the normal range according to The International Obesity Task Force BMI Cut-offs constructs were seen over the one-year interval between age (IOTF; Cole et al. 2007) where normal weight is considered to be 12 and 13 years (T2 to T3). z-BMI stability co-efficients BMI ≥ 18.6 & < 24.5 ranged from β = .89 (SE = .02) to .95 (SE = .02), with the highest stability being one year to the next. Loneliness beta sometimes” group were more likely to be in the under- weights ranged from β = .27 (SE = .03) to .47 (SE = .04), weight category and less likely than chance to be in be in with moderate to large effects (per Cohen 1988). the normal weight group and overweight weight status groups (z = 7.817, p = .012). That was not the case for girls Effects of Income Sufficiency on Weight Status and at age 13 years (z = 2.096, p = .558). Thus, it seems that the Loneliness significant difference in BMI scores between the “fre- quently lonely” versus “not lonely/sometimes lonely” Bootstrapped findings revealed no influence of earlier groups of 13-year old males was driven by the higher income sufficiency on weight status (β = .00 [SE = .02] to numbers of males from the “not lonely/sometimes lonely −.02 [SE = .02]) or loneliness (β = .05 [SE = .06] to −.11 group” in the underweight category than we would expect [SE = .07]). by chance. Taken together with the results from the ANOVA, t-tests, and chi-square analyses show no con- Weight Status Predicting Loneliness current relationships between the highest and lowest weight status and loneliness. Bootstrapped findings showed only one significant linear To examine prospective effects of weight status on effect between weight status at T2 and loneliness at T3 (β loneliness (DV) we ran a series of Hierarchical Multiple = .11 [SE = .04], p = .007), suggesting that higher weight Regressions, with predictors on the following steps: (1) status at age 12 predicted higher loneliness at age 13 years. gender, income sufficiency, and loneliness (HRAs) from the That effect is detailed in Fig. 1. There were no other sig- earlier time point, (2) z-BMI from the earlier time point, (3) nificant linear or quadratic effects of z-BMI on loneliness z-BMI squared (z-BMI ), (4) z-BMI x gender interaction, over time (See Tables S1–3 for details). 2 2 and (5) z-BMI x gender interaction. The squared ( ) term serves as a test for a quadratic relation (Cohen et al. 2003). Loneliness Predicting Weight Status Scores were centered using grand mean subtraction for loneliness, and BMI z-scores. Gender and income suffi- Bootstrapped findings showed a quadratic effect of lone- ciency were dummy coded (gender: −1 = female, and + 1 liness at T2 (age 12 years) that interacted significantly with = male; + 1 = income sufficiency, and −1 = insufficient, gender to predict z-BMI at T3 (age 13 years), β = .02, SE with the latter including the categories of insufficient and = .01, p = .007 (see Table S6 for full results). Figure 2 very insufficient) as recommended by Cohen et al. (2003). illustrates the quadratic gender-mediated effect, showing We performed bootstrapping, estimating a 95% bias- that (a) higher loneliness at age 12 years was related to corrected confidence interval for all values of interest higher z-BMI at age 13 years for girls, but (b) high lone- (1000 bootstrap sample). liness at age 12 years was related to lower z-BMI at age 13 1392 Journal of Youth and Adolescence (2018) 47:1385–1397 0.2 psychosocial consequences for both girls and boys given that the perceived ideal female body is slim, but the ideal 0.1 male body shape is muscular (Field et al. 2014). Findings 0 also suggest a significant health consequence for girls that are lonely, with increases in weight, which for girls already -0.1 overweight or obese, is a significant health risk. -0.2 The Effect of Weight Status on Loneliness Female -0.3 Male Due to the increasing importance of peer acceptance and -0.4 Low Weight Status (z-BMI) High Weight Status (z-BMI) body image concerns during early adolescence (Denneel Weight Status at T2 (12 years of age) et al. 2018; Markey 2010), and the specific peer problems Fig. 1 Weight Status at T2 (age 12 years) and Loneliness at T3 (13 encountered by youth at the extremes of weight status (Puhl years) et al. 2016), we investigated whether extremes of weight status would predict greater loneliness. Unlike previous research (Hayden-Wade et al. 2005), we did not find con- current associations between high weight status and lone- liness. We did find prospective associations between high weight status and loneliness, with male and female ado- lescents with higher weight status at age 12 years reporting increased loneliness between 12 and 13 years. Such findings that there were no within-time gender differences are con- sistent with previous research showing that male and female adolescents experience weight related criticism to a similar degree (Lawler and Nixon 2011). Further, the fact that those with higher weight status reported increasing loneliness highlights the fact that the peer context is important for understanding the social norms surrounding weight status, Female with society’s body ideals communicated and reinforced by Male peers (Jones and Crawford 2006); the fact that the pro- spective effects linking high weight status and loneliness are Low Loneliness High Loneliness only evident at ages 12 to13 years and not 10–12 years Loneliness at T2 (12 years of age) suggests a sensitive period in development when the peer Fig. 2 Slopes of the relation between loneliness T2 (age 12 years) and context becomes particularly influential in delivering mes- BMI T3 (age 13 years) as a function of gender sages about ideal body size. Future research will want to examine whether weight status exerts effects on loneliness only under specific cir- years for boys. This result suggests that loneliness may have cumstances, such as those where there is a high-level of a particular role in increasing weight for girls and reducing weight-based victimization (Juvonen et al. 2017)or an weight for boys. There were no other significant linear or internalization of the appearance ideals (Lawler and Nixon quadratic effects of loneliness on z-BMI over time (See 2011), or when internalizing problems, such as depression, Tables S4–6 for details). already exist. How loneliness explicitly links to weight- based victimization and self-esteem should also be exam- ined. In addition, time-specificinfluences of the peer con- Discussion text – in and outside school- that explain the link between high weight status and loneliness should be examined to The relationships between weight status and loneliness are determine whether there are sensitive periods during which complex. Weight status was not associated with concurrent adaptive responses to peer relationship difficulties can be feelings of loneliness, but higher weight at age 12 years most effective. predicted increased loneliness from age 12 to 13 years. Contrary to expectations, in our study low weight was Interestingly, for those who reported loneliness at age 12 not found to be concurrently associated with, or a sig- years, we observed a differential effect: weight gain for girls nificant predictor of, loneliness. The number of participants and weight loss for boys. That is likely to have detrimental who were classified as underweight was small, making it Loneliness standardized score at T3 (13 Weight Status at T3 (13 years of age) years of age) Low BMI High BMI Journal of Youth and Adolescence (2018) 47:1385–1397 1393 more difficult to gauge those effects. That said, it is also explain the prospective association between loneliness and possible that having low weight status is less socially stig- weight gain for female adolescents. matizing than being overweight or obese. It is possible that, The recommended next stage of empirical study is the due to differences in assumptions made about the volitional examination of mechanisms that explain how loneliness cause of high weight status (Puhl and Heuer 2010), it is the increases weight status among adolescent females and case that those who are underweight have fewer social decreases weight status for adolescent males. The work problems. Underweight could also be easier to disguise than detailed above may suggest that male adolescents are more overweight, leading to fewer negative social repercussions. motivated to lose weight or more aware that doing so would lead to increased social acceptance needs to be explored, but The Effect of Loneliness on Weight Status the possibility that female adolescents were engaged in emotional eating also needs to be examined. Given that In line with predictions, loneliness had a significant pro- among females, loneliness is associated with ruminative spective impact on weight status interacting significantly cognitions (Vanhalst et al. 2012), it is possible that rumi- with gender at age 12 years to predict weight status at age nation directly affects weight by influencing coping strate- 13 years. Specifically, female adolescents with higher gies (i.e., emotional eating) and planning (i.e., it reduces loneliness at age 12 years gained weight from age 12 to 13 one’s ability to stick to healthy eating intentions). Future years, while male adolescents with higher loneliness lost work should examine emotion regulation strategies, weight. Those results contribute to a growing body of including rumination, and explore whether alternative, more research indicating that loneliness affects health (Hawkley adaptive strategies for dealing with loneliness might not and Capitanio 2015), and are consistent with findings that lead to weight gain. Such work is important for informing higher loneliness leads to increased food consumption interventions focused on reducing loneliness and/or obesity. among female older adolescents (Rotenberg and Flood Future research will also want to explore how the phy- 1999). While we have not examined mechanisms linking siological effects of loneliness could influence eating loneliness and weight status, we have provided the first behavior including increasing the propensity to binge eat evidence that loneliness is directly related to increasing palatable foods and reducing one’s ability to track food weight for females upon entry into early adolescence. consumption. Such future work should take into account the The finding that loneliness reduces weight for male moderating effects of sleep and physical activity as those adolescents might suggest that loneliness serves as an factors affect weight metabolism and have been shown to be inhibitor of food consumption among those male adoles- deficient in lonely adolescents (Harris et al. 2013; Pels and cents; in contrast, loneliness may have disinhibited food Kleinert 2016b). The impact of our findings for interven- consumption for the female adolescents, a significant pro- tions to prevent obesity is clear – targeting early adoles- blem for those girls already overweight or obese females. cents, particularly females, who are lonely, could help in the The reduction in weight by male adolescents with high fight against obesity. weight status in the current sample may be demonstrable of a heightened awareness that they had to lose weight to Stability of Loneliness and Weight Status Over Time improve their social connections or were simply more motivated to do so. But, it may be the case that boys in the Results also support the stability of weight status and lone- current sample simply dealt with the stigmatization of liness over time. The relative stability of the constructs over overweight by inhibiting eating as a way to copy with time supports previous research (Pryor et al. 2011;van negative emotions. It is also possible that boys do not Dulmen and Goossens 2013). The strong stability of weight experience psychosocial effects until norms around their status and loneliness reflects the difficult task that interven- ideal male body shape (that of muscle; Field et al. 2014) tion teams face in order to affect changes in those areas of materialize during puberty. Future research should examine health. After age 13 years, the stability of weight status is those effects. likely to change due to physiological changes that accom- The weight gains we saw in girls are likely to be the pany puberty, and future work will want to examine how the outcome of engagement in emotional eating, a maladaptive onset of puberty impacts the prospective associations method of alleviating negative emotions (Haedt-Matt and between weight status and loneliness, exploring the impact Keel 2011), that they used to cope with feelings of lone- for male and female adolescents separately. Given that for liness. Goldschmidt et al. (2017) found that poor emotional females, advanced pubertal maturation is associated with awareness and limited access to adaptive emotion regulation internalizing problems, explained exclusively in terms of strategies contribute to emotional eating in adolescent environmental influences (Marceau et al. 2012), we might females, so further research will want to establish whether expect that the prospective association between weight status emotional eating and poor emotion regulation can help and loneliness is also moderated by pubertal timing for girls. 1394 Journal of Youth and Adolescence (2018) 47:1385–1397 The time interval between data collection waves in the study introduced confounds into the study. Equal time intervals varied from one to two years, but the effects were seen in the between measurement points would be preferable (Van- one-year time interval between age 12 to age 13 (T2 to T3). derWeele et al. 2011). Missing cases were dealt with That age could be a key sensitive period for peer relationship through Listwise deletion because the Data Missing Com- problems to impact weight status, future longitudinal studies pletely At Random (MCAR) tests proved significant. Our may consider one-year time intervals between waves to give chosen method is a less favored method of data cleansing a more nuanced picture of magnitude and persistence of and could have introduced bias in the sample by altering the effects over time (VanderWeele et al. 2011). standard error estimates for sub samples due to non-random data (Allison 2002). However, the use of bootstrapping Income Sufficiency and Effects on Weight Status and helps provide an assurance that the results are not spurious. Loneliness Areas for Further study In the current study, we explored the impact of income sufficiency on weight status and loneliness. We did not find Throughout this discussion we have highlighted important any effects linking recent social disadvantage, measured areas of future work. We noted the need to examine whether here in terms of income sufficiency, to high weight status. weight status exerts effects on loneliness only under specific That finding is consistent with other research (Lee et al. circumstances, including situations where there is a high- 2014) that found only poverty exposure prior to 2 years of level weight-based victimization or internalization of age had a robust association with adolescent obesity. appearance ideals. We also noted the need to further explore To our knowledge, this is the first study to examine the the gendered responses to loneliness, and how any differ- longitudinal associations between income sufficiency and ences between male and female adolescents are related to youth loneliness. We thought it might be the case that future gains or reductions in weight. children whose parents had limited income did not have In addition, in future studies, several further variables adequate time to spend with their children or have the could be controlled, including activity levels and onset of resources to support peer engagement activities, and chil- puberty. Ethnicity could not be explored as a moderator in dren from those families would experience increasing dis- the current study because the sample was not ethnically tance from parents and peers, and, thus, report loneliness. diverse. In future work, ethnicity should also be explored But, we did not find that adolescents whose families had given evidence that girls with high weight status in certain insufficient incomes as defined by Statistics Canada, ethnic groups suffer less from the negative effects of weight reported higher rates of loneliness. Further empirical work stigma than other groups (Mustillo et al. 2013). should examine income sufficiency in relation to loneliness among youth, examining whether that effect is found for children whose focus is more on parents as the main source Conclusion of support (Csikszentmihalyi and Larson 1984). The current study examined the concurrent and prospective Strengths and Limitations reciprocal relationships between weight status and loneliness, controlling for income sufficiency and gender. We found that This study has several strengths, including the large popu- both male and female adolescents with higher weight status lation sample, the robust anthropometric techniques used to reported increasing loneliness from ages 12 to 13 years, collect weight and height data, and the prospective nature of showing that society’s views of what constitutes an body the design. The current study followed children over a 3- ideal size and shape are communicated and reinforced by year period into early adolescence and allowed confirmation peers at this point in development. In addition, loneliness at of the temporal relationships between loneliness, weight age 12 years reduced weight for male adolescents from age status, and income sufficiency, and provided exploration of 12 to 13 years, and increased weight for female adolescents effects for male and female adolescents. The time period during that same period, suggesting that loneliness may serve provided a good test of the stability of the constructs during as an inhibitor of food consumption among male adolescents, a period of developmental transition. The simplicity, long- but may disinhibit food consumption for female adolescents, itudinal design, and youth sample in the current study add asignificant problem for already overweight or obese depth to our understanding of the links between weight females. Further work will want to explore the prospective status and social problems faced by young people during a effects further to determine whether the gendered effect is key period in development. specific to the current sample, and, if not, what that effect The study is not without limitations. The difference in tells us about gendered coping and peer group friendships interval lengths between data collection points could have among male and female youth. Journal of Youth and Adolescence (2018) 47:1385–1397 1395 The findings suggest early clinical attention to high weight change in a prospective community-based study of children fol- lowed up into adulthood. Archives of Pediatric Adolescent status and loneliness will be important and may have sig- Medicine, 160, 285–291. nificant effects for adolescent females. Reducing peer-related Bronfenbrenner, U. (2005). Making human beings human: Bioecolo- problems for those with high weight status will help reduce gical perspectives on human development. Thousand Oaks, CA: feelings of loneliness, but such problems reflect society’s Sage. Brylinsky, J. A., & Moore, J. C. (1994). The identification of body stigma attached to non-conformation to body ideals, making build stereotypes in young children. Journal of Research in such social norms hard to change. Thus, it will be more Personality, 28(2), 170–181. https://doi.org/10.1006/jrpe.1994. important to help adolescents develop resilient strategies of coping with peer criticism of non-ideal body size, reducing the Cacioppo, J. T., & Hawkley, L. C. (2003). Social isolation and health, with an emphasis on underlying mechanisms. Perspectives in chance that they will engage in emotional eating or experience Biology and Medicine, 46(3), S39–S52. https://doi.org/10.1353/ social stress, both of which would lead to weight gains. pbm.2003.0049. Cohen, J. (1988). Statistical power analysis for the behavioral sci- Acknowledgements This study uses data from the Quebec Long- ences (2nd ed.). Hillsdale, NJ: Erlbaum. itudinal Study of Child Development (QLSCD). Data collection for the Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied QLSCD was made possible through funding from the ministère de la multiple regression/correlation analysis for the behavioral sci- Santé et des Services Sociaux (MSSS) (Ministry of Health and Social ences. 3rd Edition New York, NY: Routledge. Services), the Lucie and André Chagnon Foundation, the ministère de Cole, T. J., Bellizzi, M. C., Flegal, K. M., & Dietz, W. H. (2000). la Famille (Ministry of Family), and the Institut de la statistique du Establishing a standard definition for child overweight and obe- Québec (the Institut). sity worldwide: International survey. British Medical Journal, 320, 1240–1243. Authors’ Contributions P.Q. conceived the study, performed statistical Cole, T. J., Flegal, K. M., Nicholls, D., & Jackson, A. A. (2007). Body analyses, interpreted the findings, drafted the manuscript, and made mass index cut offs to define thinness in children and adolescents: revisions to the manuscript based on reviewer feedback; R.H. and A. International survey. British Medical Journal, 335, 194 https:// M.E. helped to perform statistical analyses and interpret the findings, doi.org/10.1136/bmj.39238.399444.55. and contributed to the draft manuscript; J.A., M.B., and R.E.T. par- Csikszentmihalyi, M., & Larson, R. (1984). Being adolescent: Conflict ticipated in the study design, and edited the manuscript; M.B. and R.E. and growth in the teenage years. New York, NY: Basic Books. T. coordinated the original data collection. All authors read and Dallman, M. F. (2010). Stress-induced obesity and the emotional approved the final manuscript. nervous system. Trends in endocrinology and metabolics, 21(3), 159–165. https://doi.org/10.1016/j.tem.2009.10.004. Data Sharing Declaration The dataset analyzed during the current Denneel, S., Maes, M., Vanhalst, J., Bijttebier, P., & Goossens, L. study is not publicly available, but it is available from Institut de la (2018). Developmental change in loneliness and attitudes toward statistique du Québec (Information and Documentation Centre, Institut aloneness in adolescence. Journal of Youth and Adolescence, 47, de la statistique du Québec, 200, chemin Sainte-Foy, 3e étage Québec 148–161. https://doi.org/10.1007/s10964-017-0685-5. (Québec, G1R 5T4). de Onis, M., Onyango, A. W., Borghi, E., Siyam, A., Nishisa, C., & Siekmann, J. (2007). Development of a WHO growth reference for school-aged children and adolescents. Bulletin of the World Compliance with Ethical Standards Health Organization, 85, 660–667. Eschenbeck, H., Kohlmann, C. W., Dudey, S., & Schurholz, T. (2009). Conflict of Interest The authors declare that they have no conflict of Physician diagnosed obesity in German 6- to 14-year-olds. Pre- interest. valence and comorbidity of internalizing disorders, externalizing disorders, and sleep disorders. Obesity Facts, 2,67–73. Ethical Approval All procedures were in accordance with the ethical Field, A. E., Sonneville, K. R., & Crosby, R. D., et al. (2014). Pro- standards of institutional research committeesand with the 1964 Hel- spective associations of concerns about physique and the devel- sinki declaration and its later amendments or comparable ethical opment of obesity, binge drinking, and drug use among standards. This article does not contain any studies with animals. adolescent boys and young adult men. JAMA Pediatrics, 168, 34–39. https://doi.org/10.1001/jamapediatrics.2013.2915. Informed Consent Informed consent was obtained by the all indivi- Goldschmidt, A. B., Lavender, J. M., Hipwell, A. E., Stepp, S. D., & dual participants included in the study. Keenan, K. (2017). Emotion regulation and loss of control eating in community-based adolescents. Journal of Abnormal Child Open Access This article is distributed under the terms of the Creative Psychology, 45, 183–191. https://doi.org/10.1007/s10802-016- Commons Attribution 4.0 International License (http://crea 0152-x. tivecommons.org/licenses/by/4.0/), which permits unrestricted use, Goldschmidt, A. B., Sinton, M. M., Aspen, V. P., Tibbs, T. L., Stein, distribution, and reproduction in any medium, provided you give R. I., Saelens, B. E., & Wilfley, D. E (2010). Psychosocial and appropriate credit to the original author(s) and the source, provide a familial impairment among overweight youth with social pro- link to the Creative Commons license, and indicate if changes were blems. International Journal of Pediatric Obesity, 5, 428–435. made. https://doi.org/10.3109/17477160903540727. Grow, H. M. G., Cook, A. J., Arterbarn, D. E., Saelens, B. E., Drewnowski, A., & Lozano, B. E. (2010). Child obesity asso- References ciated with social disadvantage of children’s neighborhoods. Social Science and Medicine, 71, 584–591. https://doi.org/10. Allison, P. D. (2002). Missing data. Thousand Oaks, CA: Sage. 1016/j.socscimed.2010.04.018. Anderson, S. E., Cohen, P., Naumova, E. N., & Must, A. (2006). Hanlon, E. C., Tasali, E., Leproult, R., Stuhr, K. L., Doncheck, E., de Association of depression and anxiety disorders with weight Wit, H., & Van Cauter, E. (2016). Sleep restriction enhances the 1396 Journal of Youth and Adolescence (2018) 47:1385–1397 daily rhythm of circulating levels of endocannabinoid 2- and Adolescence, 41, 1111–1126. https://doi.org/10.1007/ arachidonoylglycerol. Sleep, 39, 653–664. https://doi.org/10. s10964-012-9762-y. 5665/sleep.5546. Markey, C. N. (2010). Invited commentary: Why body image is Haedt-Matt, A. A., & Keel, P. K. (2011). Revisiting the affect reg- important to adolescent development. Journal of Youth and ulation model of binge eating: A meta-analysis of studies using Adolescence, 39, 1387–1391. https://doi.org/10.1007/s10964- ecological momentary assessment. Psychological Bulletin, 137, 010-9510-0. 660–681. https://doi.org/10.1037/a0023660. Mooney, E., Farley, H., & Strugnell, C. (2009). A qualitative inves- Harris, R. A., Qualter, P., & Robinson, S. J. (2013). Loneliness tra- tigation into the opinions of adolescent females regarding their jectories from middle childhood to pre-adolescence: impact on body image concerns and dieting practices in the Republic of perceived health and sleep disturbance. Journal of Adolescence, Ireland (ROI). Appetite, 52, 485–491. https://doi.org/10.1016/j. 36, 1295–1304. https://doi.org/10.1016/j.adolescence.2012.12. appet.2008.12.012. 009. Moyer, A. E., Rodin, J., Grilo, C. M., Cummings, N., Larson, L. M., & Harrist, A. W., Swindle, T. M., Hubbs-Tait, L., Topham, G. L., Rebuffe-Scrive, M. (1994). Stress-induced cortisol response and Shriver, L. H., & Page, M. C. (2016). The social and emotional fat distribution in women. Obesity Research, 2, 255–262. https:// lives of overweight, obese, and severely obese children. Child doi.org/10.1002/j.1550-8528.1994.tb00055.x. Development, 87, 1564–1580. https://doi.org/10.1111/cdev. Mustillo, S. A., Budd, K., & Hendrix, K. (2013). Obesity, labeling, 12548. and psychological distress in late-childhood and adolescent black Hawkley, L. C., & Capitanio, J. P. (2015). Perceived social isolation, and white girls. Social Psychology Quarterly, 76, 268–289. evolutionary fitness and health outcomes: A lifespan approach. https://doi.org/10.1177/0190272513495883. Philosophical Transactions of the Royal Society B, 370, Nolen-Hoeksema, S., Wisco, B. E., & Lyubomirsky, S. (2008). 20140114 https://doi.org/10.1098/rstb.2014.0114. Rethinking rumination. Perspectives on Psychological Science, 3, Hayden-Wade, H. A., Stein, R. I., Ghaderi, A., Saelens, B. E., 400–424. https://doi.org/10.1111/j.1745-6924.2008.00088.x. Zabinski, M. F., & Wilfley, D. E. (2005). Prevalence, char- Pels, F., & Kleinert, J. (2016). Loneliness and physical activity: A acteristics, and correlates of teasing experiences among over- systematic review. International Review of Sport and Exercise weight children vs. non-overweight peers. Obesity, 13, Psychology. https://doi.org/10.1080/1750984X.2016.1177849. 1381–1392. Advance online publication. Heatherton, T. F., & Baumeister, R. F. (1991). Binge eating as escape Perlman, D., Peplau, L. A. (1981). Toward a social psychology of from self-awareness. Psychological Bulletin, 110,86–108. https:// loneliness. In S. Duck, R. Gilmour, (eds.) Personal relationships doi.org/10.1037//0033-2909.110.1.86. in disorder. (vol 3, pp 31–56). London: Academic Press. Hughes, M. E., Waite, L. J., Hawkley, L. C., & Cacioppo, J. T. (2004). Pryor, L. E., Tremblay, R. E., Boivin, M., Touchette, E., Dubois, L., A short scale for measuring loneliness in large surveys: Results Genolini, C., & Cote, S. M. (2011). Developmental trajectories of from two population-based studies. Research on Aging, 26, body mass index in early childhood and their risk factors: An 8- 655–672. year longitudinal study. Archives Pediatric and Adolescent Jones, D. C., & Crawford, J. K. (2006). The peer appearance culture Medicine, 165, 906–912. https://doi.org/10.1001/archpediatrics. during adolescence: Gender and body mass variations. Journal of 2011.153. Youth and Adolescence, 35, 257–269. https://doi.org/10.1007/ Puccio, F., Fuller-Tyszkieweicz, M., Ong, D., & Krug, I. (2016). A s10964-005-9006-5. systematic review and meta-analysis on the longitudinal rela- Juvonen, J., Lessard, L. M., Schacter, H. L., & Suchilt, L. (2017). tionship between eating pathology and depression. International Emotional implications of weight stigma across middle school: Journal of Eating Disorders, 49, 439–454. https://doi.org/10. The role of weight-based peer discrimination. Journal of Clinical 1002/eat.22506. Child and Adolescent Psychology, 46, 150–158. https://doi.org/ Puhl, R., & Brownell, K. D. (2001). Bias, discrimination, and obesity. 10.1080/15374416.2016.1188703. Obesity Research, 9, 788–805. https://doi.org/10.1038/oby.2001. Koenig, L. J., & Abrams, R. F. (1999). Adolescent loneliness and 108. adjustment: a focus on gender differences. In K. J. Rotenberg & Puhl, R. M., & Heuer, C. A. (2010). Obesity stigma: Important con- S. Hymel (Eds.), Loneliness in childhood and adolescence (pp. siderations for public health. American Journal of Public Health, 296–322). Cambridge: Cambridge University Press. 100(6), 1019–1028. https://doi.org/10.2105/AJPH.2009.159491. Lawler, M., & Nixon, E. (2011). Body dissatisfaction among adoles- Puhl, R. M., Latner, J. D., O’Brien, K., Luedicke, J., Forhan, M., & cent boys and girls: The effects if body mass, peer appearance Danielsdottir, S. (2016). Cross-national perspectives about culture and internalization of appearance ideals. Journal of Youth weight-based bullying in youth: Nature, extent and remedies. and Adolescence, 40,59–71. https://doi.org/10.1007/s10964-009- Pediatric Obesity, 11, 241–250. https://doi.org/10.1111/ijpo. 9500-2. 12051. Lee, H., Andrew, M., Gebremariam, A., Lumeng, J. C., & Lee, J. M. Puhl, R. M., & Luedicke, J. (2012). Weight-based victimization (2014). Longitudinal associations between poverty and obesity among adolescents in the school setting: Emotional reactions and from birth through adolescence. American Journal of Public coping behaviors. Journal of Youth and Adolescence, 41,27–40. Health, 104, e70–e76. https://doi.org/10.1007/s10964-011-9713-z. Little, R. J. A. (1988). A test of missing completely at random for Puhl, R. M., & Suh, Y. (2015). Stigma and eating and weight dis- multivariate data with missing values. Journal of American Sta- orders. Current Psychiatry Reports, 17,10 https://doi.org/10. tistics Association, 83, 1198e. 1007/s11920-015-0552-6. Lobstein, T., Baur, L., Uauy, R., & TaskForce, I. I. O. (2004). Obesity Qualter, P., Murphy, M., Abbott, J., Gardner, K. J., Japel, C., Boivin, in children and young people: A crisis in public health. Obesity M., Vitaro, F., & Tremblay, R. E. (2015). Developmental asso- Review, 5(Supplement 1), 4–104. https://doi.org/10.1111/j.1467- ciations between victimization and body mass index from 3 to 10 789X.2004.00133.x. years in a population sample. Aggressive Behavior, 41, 109–122. Marceau, K., Neiderhiser, J. M., Lichenstein, P., & Reiss, D. (2012). https://doi.org/10.1002/AB.21580. Genetic and environmental influences on the association between Qualter, P., Vanhalst, J., Harris, R., Van Roekel, E., Lodder, G., pubertal maturation and internalizing symptoms. Journal of Youth Bangee, M., & Verhagen, M. (2015). Loneliness across the life Journal of Youth and Adolescence (2018) 47:1385–1397 1397 span. Perspectives on Psychological Science, 10(2), 250–264. trajectories: An investigation of gene-community interplay over https://doi.org/10.1177/1745691615568999. early life course. Journal of Adolescent Health, 53, 328–334. Rankin, J., Matthews, L., Cobley, S., Han, A., Sanders, R., Wiltshire, https://doi.org/10.1016/j.jadohealth.2013.04.004. H. D., & Baker, J. S. (2016). Psychological consequences of Yang, K., & Victor, C. (2011). Age and Loneliness in 25 European childhood obesity: Psychiatric comorbidity and prevention. nations. Ageing & Society, 31,1–21. https://doi.org/10.1017/ Adolescent Health, Medicine, and Therapeutics, 7, 125–146. s0144686x1000139x. https://doi.org/10.2147/ahmt.s101631. Zawadzki, M. J., Graham, J. E., & Gerin, W. (2013). Rumination and Reitz, A. K., Zimmermann, J., Hutteman, R., Specht, J., & Neyer, F. J. anxiety mediate the effect of loneliness on depressed mood and (2014). How peers make a difference: The role of peer groups and sleep quality in college students. Health Psychology, 32, peer relationships in personality development. European Journal 212–222. https://doi.org/10.1037/a0029007. of Personality, 28, 279–288. https://doi.org/10.1002/per.1965. Rotenberg, K. J., Bharathi, C., Davies, H., & Finch, T. (2013). Bulimic symptoms and the social withdrawal syndrome. Eating Behavior, Pamela Qualter is Professor of Psychology for Education at 14, 281–284. https://doi.org/10.1016/j.eatbeh.2013.05.003. University of Manchester, Manchester Institute of Education, United Rotenberg, K. J., & Flood, D. (1999). Loneliness, dysphoria, dietary Kingdom. She leads a program of work focused on understanding the restraint, and eating behavior. International Journal of Eating predictors and outcomes of prolonged loneliness during childhood and Disorders, 25,55–64. adolescents. She is also interested in socio-emotional learning across Rotenberg, K. J., MacDonald, K. J., & King, E. V. (2004). The rela- development. tionship between loneliness and interpersonal trust during middle childhood. Journal of Genetic Psychology, 165, 233–249. Rotenberg, K. J., & Sangha, R. (2015). The relation between Bulimic Ruth Hurley is a PhD student at The School of Psychology, The symptoms and the social withdrawal syndrome during early University of Central Lancashire, United Kingdom. Her major adolescence. Eating Behaviors, 19, 177–180. https://doi.org/10. research interests include obesity, rumination, and emotional eating. 1016/j.eatbeh.2015.09.008. Sinton, M. M., Goldschmidt, A. B., Aspen, V., Theim, K. R., Stein, R. Alice M. Eccles is a PhD student at The School of Psychology, The I., Saelens, B. E., Epstein, L. H., & Wilfley, D. E. (2012). Psy- University of Central Lancashire, United Kingdom. Her major chosocial correlates of shape and weight concerns in overweight research interests include loneliness, peer relationships, and pre-adolescents. Journal of Youth and Adolescence, 41,67–75. adolescent health. https://doi.org/10.1007/s10964-011-9686-y. Turnbull, J. D., Heaslip, S., & McLeod, H. A. (2000). Pre-school children’s attitudes to fat and normal male and female stimulus Janice Abbott is a Professor of Psychology at The School of figures. International Journal of Obesity, 24, 1705–1706. https:// Psychology, The University of Central Lancashire, United Kingdom. doi.org/10.1038/sj.ijo.0801462. Her research interests are in the development and evaluation of patient- Vaillancourt, T. (2013). Do human females use indirect aggression as reported outcome measures (specifically, quality of life measurement). an intrasexual competition strategy? Philosophical Transactions of the Royal Society of London, 368(1631), 20130080 https://doi. Michel Boivin is Professor of Psychology at Université Laval, org/10.1098/rstb.2013.0080. Canada, Research Chair in Child Development and a Director of van Dulmen, M. H., & Goossens, L. (2013). Loneliness trajectories. Research Unit on Children’s Psychosocial Maladjustment (GRIP). He Journal of Adolescence, 36, 1247–1249. https://doi.org/10.1016/ leads a program of research on biological, psychological, and social j.adolescence.2013.08.001. components of child development, a program that is mainly anchored VanderWeele, T. J., Hawkley, L. C., Thisted, R. A., & Cacioppo, J. T. to population-based longitudinal studies of children, including the (2011). A marginal structural model analysis for loneliness: Quebec Longitudinal Study of Child Development and the Quebec Implications for intervention trials and clinical practice. Journal of Consulting and Clinical Psychology, 79(2), 225–235. https:// Newborn Twin Study. doi.org/10.1037/a002261. Vanhalst, J., Luyckx, K., Raes, F., & Goossens, L. (2012). Loneliness Richard E. Tremblay is Professor of Pediatrics, Psychiatry, and and depressive symptoms: The mediating and moderating role of Psychology at The University of Montreal and holds the Canada uncontrollable ruminative thoughts. The Journal of Psychology, Research Chair in Child Development. He is also a Director of the 146, 259–276. https://doi.org/10.1080/00223980.2011.555433. Research Unit on Psychosocial Maladjustment (GRIP), a Wang, J., Iannotti, R., & Luk, J. (2010). Bullying victimization among multidisciplinary research center. His research interests include child underweight and overweight U.S. youth: Differential associations development, education, prevention, family, parent-child relations, for boys and girls. Journal of Adolescent Health, 47,99–101. child psychopathology, aggressive behavior, delinquency, and https://doi.org/10.1016/j.jadohealth.2009.12.007. substance abuse. Wickrama, K. A. S., O’Neal, C. W., & Lee, T. K. (2013). Early community context, genes, and youth’s body mass index http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Youth and Adolescence Springer Journals

Reciprocal Prospective Relationships Between Loneliness and Weight Status in Late Childhood and Early Adolescence

Free
13 pages

Loading next page...
 
/lp/springer_journal/reciprocal-prospective-relationships-between-loneliness-and-weight-7CIruU5DiH
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Psychology; Child and School Psychology; Clinical Psychology; Health Psychology; Law and Psychology; History of Psychology; Psychology, general
ISSN
0047-2891
eISSN
1573-6601
D.O.I.
10.1007/s10964-018-0867-9
Publisher site
See Article on Publisher Site

Abstract

Adolescents who do not conform to weight ideals are vulnerable to disapproval and victimization from peers in school. But, missing from the literature is a prospective examination of weight status and feelings of loneliness that might come from those experiences. Using data from the Québec Longitudinal Study of Child Development, we filled that gap by examining the prospective associations between loneliness and weight status when the sample was aged 10–13 years. At ages 10, 12, and 13 years, 1042 youth (572 females; 92% from French speaking homes) reported on their loneliness and were weighed and measured. Family income sufficiency was included in our analyses given its relationship with weight status, but also its possible link with loneliness during early adolescence. The findings showed that (1) weight status and loneliness were not associated concurrently; (2) weight status predicted increases in loneliness from ages 12 to 13 years; and (3) loneliness predicted increases in weight from ages 12 to 13 years among female adolescents, but weight loss among male adolescents. The fact that loneliness was involved in weight gain for females suggests that interventions focused on reducing loneliness and increasing connection with peers during early adolescence could help in reducing obesity. ● ● ● ● ● Keywords Loneliness Body mass index Coping Income sufficiency Socioeconomic status Obesity Introduction et al. 2016), heightened loneliness among 10–14 years olds (Hayden-Wade et al. 2005) and weight concerns in pre- Social relationships and weight are prevalent concerns adolescents (Sinton et al. 2012). However, the prospective during adolescence (Denneel et al. 2018; Markey 2010). relationship between obesity and loneliness has yet to be Adolescents are vulnerable to peer disapproval of body size examined. (Lawler and Nixon 2011), feelings of loneliness (Qualter Building on longitudinal work that not only indicates et al. 2015) and they are driven by a need to “fitin” (Reitz that obesity predicts depression, but that depression also et al. 2014). Given the stigma associated with being predicts obesity (Goldschmidt et al. 2010), we hypo- overweight or obese, peer disapproval is high, creating thesized a bidirectional relationship between loneliness and negative social consequences (Harrist et al. 2016; Puhl high weight status. Conversely, feelings of loneliness are reported by those with eating disorders characterized by low body weight (Puhl and Suh 2015) and a meta-analysis indicated that low weight and depressive symptoms are Electronic supplementary material The online version of this article (https://doi.org/10.1007/s10964-018-0867-9) contains supplementary bidirectionally-related over time (Puccio et al. 2016). material, which is available to authorized users. Therefore, the current work with adolescents pro- spectively examines the relationship between (a) loneliness * Pamela Qualter and high weight status and (b) loneliness and low body pamela.qualter@manchester.ac.uk weight. Manchester Institute of Education, University of Manchester, Oxford Road, Manchester M13 9PL, UK The Social Context of Weight Status The School of Psychology, The University of Central Lancashire, Preston, UK The bio-ecological framework highlights two contexts Université Laval, Québec, QC G1V 0A6, Canada important for adolescent health because they influence The University of Montreal, Québec, Canada beliefs and behavior: (1) the immediate peer group, and (2) 1234567890();,: 1234567890();,: 1386 Journal of Youth and Adolescence (2018) 47:1385–1397 the wider social context within which the individual and (Rankin et al. 2016; Puccio et al. 2016). Findings show that peers live (Bronfenbrenner 2005). Social exclusion by peers obese 12–14 year olds have increased chance of developing in school and the accompanying feelings of loneliness depression (Eschenbeck et al. 2009), but also that depres- during adolescence are recognized as significant influences sive symptoms predict weight gain (Goldschmidt et al. of adolescent health (Hawkley and Capitanio 2015). But, 2010). The same bi-directional effects have been found for social exclusion based on weight status needs to also be those with low weight status too, with low weight predict- understood within the wider social context, where social ing depressive symptoms, but also depressive symptoms norms of the ideal body size create stigma associated with predicting decreases in weight (Puccio et al. 2016). The non-ideal weight, influencing adolescents to criticize their effects appear to be more pronounced for adolescent girls peers’ appearance (Lawler and Nixon 2011) and tease them compared to boys (Anderson et al. 2006). for non-conformation to weight ideals (Mooney et al. 2009). Mechanisms linking depression with obesity (Gold- While male and female adolescents are often criticized schmidt et al. 2017) and low weight (Puccio et al. 2016) about their appearance by their peers to a similar degree during adolescents have been examined. That evidence (Lawler and Nixon 2011), the social norms of ideal weight shows that distress caused from the feelings of shame and are different for males and females because gender stereo- guilt of not conforming to the body and weight ideal con- types of the socio-cultural ideal of beauty emphasize thin- tribute to obesity and low weight being linked over time ness for women and female adolescents (Puhl and Brownell with depression. Goldschmidt et al. also showed that 2001). Such findings suggest that female adolescents may depressive symptoms predict engagement in emotional be more vulnerable to the social context of appearance than eating as a way to alleviate distress. male adolescents, receiving greater pressure from peers to conform to the socio-cultural ideals. Empirical evidence Weight Status and Loneliness shows that females are particularly vulnerable to the nega- tive social effects of high weight status, experiencing more In contrast to the work exploring weight status and rejection and victimization from the peer group than boys depressive symptoms during adolescence, there is a paucity after the age of seven years (Qualter et al. 2015). Thus, of research examining concurrent and prospective rela- gender-based social norms are important to consider when tionships between weight status and loneliness at that same examining the interaction between weight status and lone- development period. Loneliness is the negative feeling that liness because females may be particularly vulnerable to the occurs when a person does not perceive their social rela- negative social effects of not conforming to weight status tionships to be as satisfying as they would like (Perlman ideals, experiencing, as a result, more loneliness than boys. and Peplau 1981), sharing one common symptom with Other society level influences are also important to depression (negative affect). But, loneliness is determined, consider in the current work. Socioeconomic adversity specifically, by a negative emotional response to a lack of during childhood and adolescence is linked to develop- close affiliative ties to peers during adolescence; depres- mental processes: low income is a known risk factor for sion, in contrast, is attributed to a broader range of obesity earlier in development (Grow et al. 2010) because it causes, including determinants other than deficient social leads to stress responses that exacerbate metabolic pro- relations (i.e., the individual’s genetic makeup, neurological cesses, leading to increased weight status (Wickrama et al. disorders, psychological dysfunctions; Koenig and Abrams 2013). Further, there is reason to think that low income may 1999). Thus, an exploration of the prospective association also be linked to loneliness. To our knowledge, there is between weight status and loneliness is important given currently no research on socioeconomic status and lone- that loneliness specifically taps distress associated with liness during adolescence, but parents who have limited peer group problems, which are hypothesized to be a sig- resources and income may not find adequate time to spend nificant part of the puzzle linking weight status with with their children, contributing to increasing distance depression. between parents and their children and child negligence; Research exploring the mechanisms linking weight status they also may not have the financial resources to ensure and depression discusses (1) how peer problems are a cat- their child’s engagement in specific peer group activities alyst for depressive symptoms among adolescents with the that cost money. For those reasons, we include family highest and lowest weight statuses, but also (2) how coping income sufficiency in our analyses. with emotional distress can lead to emotional eating and increases in weight status. Based on the fact that loneliness Weight Status and Depressive Symptoms includes the same negative affect as depression, we hypo- thesize a prospective bidirectional association between Being at the extremes of weight status (overweight/obese or loneliness and weight status, which we discuss further underweight) is associated with depressive symptoms below. Journal of Youth and Adolescence (2018) 47:1385–1397 1387 Weight Status Predicting Loneliness empirical studies. Consistent with the affect-regulation model of binge-eating (Heatherton and Baumeister 1991), It is clear, from empirical research, that there is stigma attached evidence shows that induced loneliness leads to increased to high weight status. Empirical evidence shows that, in eating as a way to alleviate distress (Rotenberg and Flood western society, the high degree of social stigma attached to 1999). But, loneliness has also been shown to increase obesity/overweight is evident as early as pre-school (Turnbull stress hormones such as cortisol that come from heightened et al. 2000), with increasing negative ratings for “chubby” physiological stress during episodes of loneliness body types by children aged 5–10 years (Brylinsky and Moore (Cacioppo and Hawkley 2003), and those stress hormones 1994) and peer social rejection of obese children at ages 6–7 affect fat storage and transportation with the body (Dallman years(Harristetal. 2016). Thosenegativesocial circumstances 2010; Moyer et al. 1994). continue into adolescence, with obese and overweight ado- In addition to the direct weight-related effects of stress lescents being regularly stigmatized, socially excluded, and hormones, a combination of perceived stress, disturbed victimized by peers (Puhl and Luedicke 2012;Puhletal. sleep, and cognitive rumination resulting from loneliness 2016). Thus, high weight status is associated with poorer peer (Zawadzki et al. 2013) could affect eating behavior by relationships, with overweight 10–14 year olds reporting impairing one’s ability to address problems, leading to higher levels of loneliness (Hayden-Wade et al. 2005). Given greater use of passive coping strategies, such as emotional that evidence, one could posit that higher weight status would eating (Nolen-Hoeksema et al. 2008). Those loneliness- result in less satisfaction with social relationships over time related stressors are also argued to increase our seeking of i.e., greater feelings of perceived loneliness. To date, however, easy learned-rewards such as high fat and high sugar foods the prospective link between higher weight status and feelings due to stimulation of the motivation and reward circuits in of loneliness has not been examined. the brain (Dallman 2010; Hanlon et al. 2016), making it In addition to predicting that higher weight status could easier to gain weight and harder to lose it. result in greater loneliness, it is also possible that those with The work represented above supports the idea that low weight status will report increasing loneliness over loneliness may lead to increases in weight status, but the time. Empirical work by Wang et al. (2010) shows that male prospective link has not yet been examined. The current and female adolescents with high weight status are often study examines whether loneliness predicts increases in bullied by their peers, but underweight girls are also often weight during late childhood/early adolescence. victims. That aggression, and the accompanying feelings of shame and social isolation from peers, is thought to be a consequence of intrasexual competition that is promoted by The Current Study society’s emphasis on thinness among females (Vaillan- court 2013). Rotenberg et al. (2013) and Rotenberg and An examination of the reciprocal prospective relationships Sangha (2015) have also shown that adolescents with low between loneliness and weight status among adolescence is weight status as a result of eating pathologies report higher important if we want to offer effective intervention solu- levels of loneliness compared to their peers, a consequence tions for loneliness and obesity among youth. In the current of stigmatization and alienated from the peer group. study, feelings of loneliness and weight status are explored Given the work detailed above, there is a need to in a population sample of Canadian 10–13 year olds, and examine whether both high and low weight status predict we examine whether high and low weight status put ado- increasing loneliness over the adolescent years. We pos- lescents at risk of later loneliness, but also whether lone- tulate that there will be concurrent and prospective asso- liness scores predict increasing weight status. The findings ciations between weight status and loneliness, and that could have important implications for interventions for those relationships will be curvilinear in nature, such that obese or underweight youth, and for adolescents who report individuals with either the highest or lowest weight sta- loneliness. Guided by the bioecological framework, we also tuses will feel more loneliness during adolescence com- seek to examine gender and income sufficiency as mod- pared to their normal weight peers. The current study will erators of the prospective relationships between weight examine the prospective curvilinear associations between status and loneliness. weight status and loneliness. Loneliness as a Predictor of Increasing Weight Method Status This study utilized data from the Québec Longitudinal Among adults, loneliness has been tentatively implicated as Study of Child Development (QLSCD), a large on-going a risk factor for increased weight, but there are few study which has tracked the health and wellbeing of a 1388 Journal of Youth and Adolescence (2018) 47:1385–1397 Table 1 Characteristics of the Time point Age of participants in months n = Sample at the Start of the (SD) Quebec Longitudinal Study of Child Development (QLSCD) Start of QLSCD survey in 1998 4.5 (0.55) 2120 survey and at T1–T3 of the T1 current study 121.70 (3.10) 1042 current study T2 current study 145.60 (3.05) 1042 T3 current study 157.60 (3.12) 1042 Demographics taken at start of QLSCD for the current sample compared to original sample Sample chosen for QLSCD Sample in current study Males (%) 51 45 French-speaking families (%) 81 92 Mother’s age in years 29 31 Father’s age in years 31 33 Mothers did not hold a high school degree (%) 17 15 Fathers did not hold a high school degree (%) 20 16 Mother had a university degree (%) 28 32 Father had a university degree (%) 26 27 Income Sufficiency at normal levels 81 88 Households headed by single parent (%) 7 5 Statistics Canada’s low-income cut-offs (LICOs) were used to categorize the families of participants on income sufficiency. LICOs are income thresholds at which a family would typically spend 20% more of its income than the average family on the necessities of food, shelter, and clothing. Families are classified as having “sufficient income” when the household income is above the low-income threshold determined by Statistics Canada in any given year. When income is between 60 and 90% of the low-income threshold, households are classified as having “insufficient income”; income levels below 60% of the low-income threshold are considered as “very insufficient” QLSCD Quebec longitudinal study of child development random sample of Quebec infants on a range of measures T1–T3 of the current study. Participants with and without since they were 5 months old (see the study website all data for the period of the current study were compared http://www.iamillbe.stat.gouv.qc.ca/default_an.htm for fur- using Little’s missing completely at random test (Little ther details). The representative sample is made up of 1988) to determine whether data imputation would be children born throughout 1998 in the Canadian province of possible. That comparison resulted in a significant chi- Québec (total population over 7 million, with approximately square value, Χ (18) = 35.246, p = .009, suggesting that 70,000 newborns per year). A total of 2,940 infants were missing values could not be dealt with using data imputa- selected for QLSCD through a region-based stratified tion methods. Thus, we used listwise deletion of cases, sampling design, of which 2120 infants (48.8% girls) took analyzing data from 1042 participants (572 females, 470 part, with parents providing informed consent in 1998. males) for whom complete loneliness, height, and weight Twins and children with major diseases at birth were not data were available at all three time points . Table 1 shows part of the study. In the current study, four waves of data how the final sample in the current study compares to the were used from successive data collections when the sample original QLSCD sample that had been chosen as a repre- of children was aged 10, 12, and 13 years. We refer to these sentative sample of children in Québec in 1998. The table time points as T1, T2, and T3 respectively of the current shows that, while there were fewer males in the current study. The QLSCD was approved by the Health Research Ethics Committees of the Québec Statistics Institute and the Loneliness, weight, and height data were also available for 926 of University of Montreal. the 1042 children at a further time point when the sample was aged 15 years. However, the measure of loneliness was limited to a single item, with response items that did not align with that for the earlier lone- Participants liness items; height and weight data were self-reported. In addition, there were so few participants who were classified as underweight (N Loneliness, weight, and height data were collected from = 20 females; 9 males) or obese (N = 9 females; 5 males) that it would 1259 children/early adolescents (667 females, 592 males) at not make the analyses viable. Journal of Youth and Adolescence (2018) 47:1385–1397 1389 sample than in the original sample, the reduced sample recommendations from The International Obesity Task taking part in the current study was representative of the Force BMI (IOTF; Cole et al. 2007). Those recommenda- same children living in the province of Québec in 1998. tions classified participants into the following weight cate- gories at each time point: thin grade 3 (BMI ≤ 16), thin Measures grade 2 (BMI ≤ 17 & > 16), thin grade 1 (BMI ≤ 18.5 & > 17), overweight (BMI ≥ 25 & < 30), obese (BMI ≥ 30) Loneliness morbid obesity (BMI > 30), and normal weight (BMI ≥ 18.6 & ≤ 24.5). Those IOTF cut-offs are recommended for Loneliness was measured using the Loneliness and Social international research and comparison (Lobstein et al. Satisfaction questionnaire developed by Rotenberg et al. 2004), and are used here to create weight status groups for (2004). This three-item measure is similar to the 3-item each time point, enabling us to explore whether different short form of the UCLA (University of California, Los weight categories reported higher levels of loneliness at Angeles) Loneliness Scale (Hughes et al. 2004), but the each time point. word “isolated” was simplified to “alone” in item 3. Items asked the extent to which, in the last two weeks, participants Income sufficiency had felt (1) “they had people they could talk to”, (2) “left out of things” and, (3) “alone”. Item 1 was reverse coded so Sufficiency of income was determined by low-income cut- that higher scores represented higher feeling of loneliness. offs set by Statistics Canada in any given year. It takes into Participants answered how they best described those feel- account household income, the size of the household, and ings (1 = never, 2 = sometimes, 3 = always), with possible the size of the residence area. Statistics Canada’s low- total scores ranging from 3 to 9. Cronbach’s alpha for the income cut-offs (LICOs) are income thresholds at which a loneliness measure was .66, .68, and .74 at T1, T2, and T3 family would typically spend 20% more of its income than respectively. Total scores on the loneliness scale were used the average family on the necessities of food, shelter, and in our regression analyses looking at the prospective asso- clothing (https://www.statcan.gc.ca/pub/75f0002m/2012002/ ciations between loneliness and weight status, but we also lico-sfr-eng.htm). Families are classified as having “sufficient created loneliness groups to examine whether children who income” when the household income is above the low- reported higher levels of loneliness at any given time also income threshold determined by Statistics Canada. When had higher weight status scores (higher BMI) at that same income is between 60 and 90% of the low-income thresh- time point. Following Yang and Victor (2011) we created a old, households are classified as having “insufficient “frequently lonely” group of participants who scored 6 or income”; income levels below 60% of the low-income above on the loneliness scale; all remaining youth were threshold are considered as “very insufficient”. Although grouped into a “not lonely/occasionally lonely” category. LICOs are widely used, they do not measure poverty. Those groups were created for the purpose of making Unlike the US, Canada does not have a measure of poverty. comparisons on BMI at each time point. For example, according to Statistics Canada (https://www. statcan.gc.ca/pub/75f0002m/75f0002m2010003-eng.htm) Weight status low-income thresholds are different for a family living in a rural area compared to similar families living in large cities. Trained research assistants weighed and measured the par- ticipants when they were wearing lightweight clothing and Analyses Plan no shoes. Two measurements were taken, and, if they varied by more than 0.5 cm for height or 0.2 kg for weight, a third First, using ANOVA, we examined whether there were measurement was taken. Where multiple measures were differences between people in the different weight cate- available, we used the average of each measure to calculate gories on loneliness at each time point. Second, using T- BMI (BMI = kg/m2). The computation of the participants’ tests, we examined differences in BMI between groups of BMI was followed by the creation of a BMI z-score using youth identified as either “frequently lonely” or “not lonely/ respondents’ BMI, self-reported age, gender, and the occasionally lonely”. We conducted all analyses separately external reference sample from WHO (Cole et al. 2000;de by gender. Third, we conducted a series of chi-square tests Onis et al. 2007). Those BMI z-scores (referred to here as z- to clarify the associations between loneliness and weight BMI) were used in our regression analyses, exploring the status, exploring for males and females separately, mem- prospective association between weight status (z-BMI) and bership of any given loneliness group by weight category at loneliness. As well as create a z-BMI score for each parti- each time point. Those chi-square analyses helped us to cipant, we also classified each of them as underweight, understand whether adolescents who were “frequently overweight, obese, or normal weight, following lonely” at any time point were more likely than chance to 1390 Journal of Youth and Adolescence (2018) 47:1385–1397 Table 2 Mean (and standard deviations) for loneliness in each weight category at each time point for females and males Time (age) Underweight Normal range Overweight Obese BMI < 18.5 18.51–24.99 BMI ≥ 25 BMI ≥ 30 Females Males Females Males Females Males Females Males T1 (10 year) 3.64 (.93) 43.46 (.95) 3.82 (1.19) 3.88 (1.23) 3.97 (1.33) 3.86 (1.29) 4.03 (1.27) 3.72 (.96) N = 1042 36 26 407 325 99 90 30 29 T2 (12 year) 3.71 (.86) 3.28 (.61) 3.76 (1.19) 3.81 (1.25) 3.90 (1.17) 3.73 (1.09) 3.97 (1.17) 3.58 (.97) N = 1042 35 25 384 311 116 98 37 36 T3 (13 year) 3.80 (1.10) 3.93 (1.51) 3.83 (1.31) 3.92 (1.29) 3.85 (1.14) 4.06 (1.43) 3.84 (1.26) 3.82 (1.29) N = 1042 30 28 384 299 120 105 38 38 Notes: Sample participants were categorized using the following recommendations from The International Obesity Task Force BMI (IOTF; Cole et al. 2007): thin grade III (BMI ≤ 16), thin grade II (BMI ≤ 17 & > 16), thin grade I (BMI ≤ 18.5 & > 17), overweight (BMI ≥ 25 & < 30), obese (BMI ≥ 30), morbid obesity (BMI > 30), and normal weight (BMI ≥ 18.6 & < 24.5). Underweight = Grade I, II, III were combined due to low numbers in Grades II and III; Obese (obese and morbid obese were combined due to low numbers in the morbid obese category); Possible loneliness scores ranged from 3 to 9. N = 1042 (Female = 572; Male = 470). ANOVAs revealed no differences between males and females on feelings of loneliness at each time point also be members of the obese or underweight weight F (3, 466) = 1.835, p = .140; and T3 (age 13 years) groups. Together, those sets of analyses enabled us to females, F (3, 568) = .014, p = .998, and males, F (3, 466) examine concurrent relationships between loneliness and = .393, p = .758. Thus, in the current population sample, it weight status. seems weight status is not associated with concurrent Fourth, we examined prospective linear and curvilinear reports of higher loneliness during adolescence for males or relationships between (1) z-BMI and loneliness, and (2) females; those with high or low weight status did not report loneliness and z-BMI using Hierarchical Regression Ana- higher levels of loneliness compared to normal weight lyses (HRAs), controlling for gender and income suffi- category to report feelings of loneliness (Table 2). ciency. Curvilinear analyses were used to determine the The next set of analyses examined whether adolescents exact associations between weight status at one time classified as “frequently lonely” were different to peers point and loneliness at the following data collection categorized as “not lonely/sometimes lonely” on weight waves, and loneliness at one time point and weight status at status at each of the time points. We ran a series of inde- the other data collection waves. Based on our findings pendent t-tests, separately for males and females and found of quadratic effects between z-BMI and loneliness, no difference between the “frequently lonely” and “not Structural Equation Modelling was not an appropriate lonely/sometimes lonely” groups at age 10 years (T1: analytic tool and the HRAs are presented as our final sta- females, t (74.78) = 1.88, p = .064; males, t (465) = 1.12, p tistical analyses. = .264) or age 12 years (T2: females, t (570) = 1.26, p = .210; males, t (467) = .03, p = .974). At age 13 years (T3), there was a difference between the lonely groups for Results males (t (81.38) = 3.04, p = .003), but not females (t (77.77) = 1.53, p = .131). Findings showed that male ado- We examined differences in mean loneliness for young lescents aged 13 years in the “frequently lonely” group had people in different weight categories using ANOVAs, a higher weight status than their male peers in the “not looking at each time point and males and females sepa- lonely/sometimes lonely” group (Table 3). rately. Because there were so few participants in the three Given that the mean BMI scores at T3 for males in the grades of thinness as defined by (Cole et al. 2007), we “frequently lonely” group would be considered within the merged those groups to create one group that we defined as normal range of BMI scores according to The International “underweight” (BMI < 18.5). We also merged the obese and Obesity Task Force BMI cut-offs (Cole et al. 2007; normal morbidly obese groups given that there were so few parti- weight = BMI ≥ 18.6 & ≤ 24.5), we decided to explore the cipants classified as morbidly obese. concurrent association between loneliness and weight status The one-way ANOVAs investigating loneliness by at T3 further. We conducted two Fisher-Freeman-Halton weight category revealed no significant differences at any of tests, one for males and one for females. Those analyses the three time periods: T1 (age 10 years) females, F (3, 568) showed that, at age 13 years (T3), males classified as = .994, p = .395, males, F (3, 466) = 1.057, p = .367; T2 “frequently lonely” were more likely than chance to be in (age 12 years) females, F (3, 568) = .741, p = .528, males, the normal weight category, while boys in the “not lonely/ Journal of Youth and Adolescence (2018) 47:1385–1397 1391 Table 3 Mean (and standard deviations) for BMI by loneliness group A further set of HRAs was conducted to examine the at each time point for females and males longitudinal over-time effects of loneliness on BMI. The Time Frequently lonely Not lonely/Sometimes lonely regressions followed the pattern and procedure outlined Females Males Females Males above, with z-BMI at each time point as the independent variable, and loneliness and z-BMI from earlier time points T1 (10 year) 19.26 (4.17) 17.76 (2.32) 18.28 (3.11) 18.25 (2.06) as predictors. Any two-way interactions between z-BMI x N = 1042 66 54 506 416 T2 (12 year) 20.63 (3.43) 19.86 (3.92) 19.98 (3.88) 19.84 (3.76) Gender were further examined by testing for the linear and N = 1042 60 49 512 421 quadratic (curvilinear) relations on the measure for each T3 (13 year) 21.82 (5.39) 22.52 (5.11) 20.80 (3.84) 20.56 (3.79) gender separately. Key information from the HRAs are N = 1042 69 69 503 401 detailed in the manuscript text, with tables (Tables S1-S6) detailing all HRA information included as on-line supple- Notes: Those in the “Frequently Lonely” group scored 6 or above on the loneliness scale; those scoring between 3 and 5 were classified as mentary information. “Not Lonely/Sometimes Lonely”;N = 1042 (Female = 572; Male = 470); T-tests showed no differences on BMI between same sex peers Stability of Weight and Loneliness Over Time in the “Frequently Lonely” and “Not Lonely/Sometimes Lonely” groups at ages 10 and 12 years. At age 13 years, males in the “Frequently Lonely” group scored significantly higher on BMI As anticipated, over time the strongest predictor of z-BMI compared to their same sex peers in the “Not Lonely/Sometimes and loneliness were previous measures of the same con- Lonely” group; there were no significant differences between females. struct, confirming stability. The strongest effects for both All BMI mean scores would be considered in the normal range according to The International Obesity Task Force BMI Cut-offs constructs were seen over the one-year interval between age (IOTF; Cole et al. 2007) where normal weight is considered to be 12 and 13 years (T2 to T3). z-BMI stability co-efficients BMI ≥ 18.6 & < 24.5 ranged from β = .89 (SE = .02) to .95 (SE = .02), with the highest stability being one year to the next. Loneliness beta sometimes” group were more likely to be in the under- weights ranged from β = .27 (SE = .03) to .47 (SE = .04), weight category and less likely than chance to be in be in with moderate to large effects (per Cohen 1988). the normal weight group and overweight weight status groups (z = 7.817, p = .012). That was not the case for girls Effects of Income Sufficiency on Weight Status and at age 13 years (z = 2.096, p = .558). Thus, it seems that the Loneliness significant difference in BMI scores between the “fre- quently lonely” versus “not lonely/sometimes lonely” Bootstrapped findings revealed no influence of earlier groups of 13-year old males was driven by the higher income sufficiency on weight status (β = .00 [SE = .02] to numbers of males from the “not lonely/sometimes lonely −.02 [SE = .02]) or loneliness (β = .05 [SE = .06] to −.11 group” in the underweight category than we would expect [SE = .07]). by chance. Taken together with the results from the ANOVA, t-tests, and chi-square analyses show no con- Weight Status Predicting Loneliness current relationships between the highest and lowest weight status and loneliness. Bootstrapped findings showed only one significant linear To examine prospective effects of weight status on effect between weight status at T2 and loneliness at T3 (β loneliness (DV) we ran a series of Hierarchical Multiple = .11 [SE = .04], p = .007), suggesting that higher weight Regressions, with predictors on the following steps: (1) status at age 12 predicted higher loneliness at age 13 years. gender, income sufficiency, and loneliness (HRAs) from the That effect is detailed in Fig. 1. There were no other sig- earlier time point, (2) z-BMI from the earlier time point, (3) nificant linear or quadratic effects of z-BMI on loneliness z-BMI squared (z-BMI ), (4) z-BMI x gender interaction, over time (See Tables S1–3 for details). 2 2 and (5) z-BMI x gender interaction. The squared ( ) term serves as a test for a quadratic relation (Cohen et al. 2003). Loneliness Predicting Weight Status Scores were centered using grand mean subtraction for loneliness, and BMI z-scores. Gender and income suffi- Bootstrapped findings showed a quadratic effect of lone- ciency were dummy coded (gender: −1 = female, and + 1 liness at T2 (age 12 years) that interacted significantly with = male; + 1 = income sufficiency, and −1 = insufficient, gender to predict z-BMI at T3 (age 13 years), β = .02, SE with the latter including the categories of insufficient and = .01, p = .007 (see Table S6 for full results). Figure 2 very insufficient) as recommended by Cohen et al. (2003). illustrates the quadratic gender-mediated effect, showing We performed bootstrapping, estimating a 95% bias- that (a) higher loneliness at age 12 years was related to corrected confidence interval for all values of interest higher z-BMI at age 13 years for girls, but (b) high lone- (1000 bootstrap sample). liness at age 12 years was related to lower z-BMI at age 13 1392 Journal of Youth and Adolescence (2018) 47:1385–1397 0.2 psychosocial consequences for both girls and boys given that the perceived ideal female body is slim, but the ideal 0.1 male body shape is muscular (Field et al. 2014). Findings 0 also suggest a significant health consequence for girls that are lonely, with increases in weight, which for girls already -0.1 overweight or obese, is a significant health risk. -0.2 The Effect of Weight Status on Loneliness Female -0.3 Male Due to the increasing importance of peer acceptance and -0.4 Low Weight Status (z-BMI) High Weight Status (z-BMI) body image concerns during early adolescence (Denneel Weight Status at T2 (12 years of age) et al. 2018; Markey 2010), and the specific peer problems Fig. 1 Weight Status at T2 (age 12 years) and Loneliness at T3 (13 encountered by youth at the extremes of weight status (Puhl years) et al. 2016), we investigated whether extremes of weight status would predict greater loneliness. Unlike previous research (Hayden-Wade et al. 2005), we did not find con- current associations between high weight status and lone- liness. We did find prospective associations between high weight status and loneliness, with male and female ado- lescents with higher weight status at age 12 years reporting increased loneliness between 12 and 13 years. Such findings that there were no within-time gender differences are con- sistent with previous research showing that male and female adolescents experience weight related criticism to a similar degree (Lawler and Nixon 2011). Further, the fact that those with higher weight status reported increasing loneliness highlights the fact that the peer context is important for understanding the social norms surrounding weight status, Female with society’s body ideals communicated and reinforced by Male peers (Jones and Crawford 2006); the fact that the pro- spective effects linking high weight status and loneliness are Low Loneliness High Loneliness only evident at ages 12 to13 years and not 10–12 years Loneliness at T2 (12 years of age) suggests a sensitive period in development when the peer Fig. 2 Slopes of the relation between loneliness T2 (age 12 years) and context becomes particularly influential in delivering mes- BMI T3 (age 13 years) as a function of gender sages about ideal body size. Future research will want to examine whether weight status exerts effects on loneliness only under specific cir- years for boys. This result suggests that loneliness may have cumstances, such as those where there is a high-level of a particular role in increasing weight for girls and reducing weight-based victimization (Juvonen et al. 2017)or an weight for boys. There were no other significant linear or internalization of the appearance ideals (Lawler and Nixon quadratic effects of loneliness on z-BMI over time (See 2011), or when internalizing problems, such as depression, Tables S4–6 for details). already exist. How loneliness explicitly links to weight- based victimization and self-esteem should also be exam- ined. In addition, time-specificinfluences of the peer con- Discussion text – in and outside school- that explain the link between high weight status and loneliness should be examined to The relationships between weight status and loneliness are determine whether there are sensitive periods during which complex. Weight status was not associated with concurrent adaptive responses to peer relationship difficulties can be feelings of loneliness, but higher weight at age 12 years most effective. predicted increased loneliness from age 12 to 13 years. Contrary to expectations, in our study low weight was Interestingly, for those who reported loneliness at age 12 not found to be concurrently associated with, or a sig- years, we observed a differential effect: weight gain for girls nificant predictor of, loneliness. The number of participants and weight loss for boys. That is likely to have detrimental who were classified as underweight was small, making it Loneliness standardized score at T3 (13 Weight Status at T3 (13 years of age) years of age) Low BMI High BMI Journal of Youth and Adolescence (2018) 47:1385–1397 1393 more difficult to gauge those effects. That said, it is also explain the prospective association between loneliness and possible that having low weight status is less socially stig- weight gain for female adolescents. matizing than being overweight or obese. It is possible that, The recommended next stage of empirical study is the due to differences in assumptions made about the volitional examination of mechanisms that explain how loneliness cause of high weight status (Puhl and Heuer 2010), it is the increases weight status among adolescent females and case that those who are underweight have fewer social decreases weight status for adolescent males. The work problems. Underweight could also be easier to disguise than detailed above may suggest that male adolescents are more overweight, leading to fewer negative social repercussions. motivated to lose weight or more aware that doing so would lead to increased social acceptance needs to be explored, but The Effect of Loneliness on Weight Status the possibility that female adolescents were engaged in emotional eating also needs to be examined. Given that In line with predictions, loneliness had a significant pro- among females, loneliness is associated with ruminative spective impact on weight status interacting significantly cognitions (Vanhalst et al. 2012), it is possible that rumi- with gender at age 12 years to predict weight status at age nation directly affects weight by influencing coping strate- 13 years. Specifically, female adolescents with higher gies (i.e., emotional eating) and planning (i.e., it reduces loneliness at age 12 years gained weight from age 12 to 13 one’s ability to stick to healthy eating intentions). Future years, while male adolescents with higher loneliness lost work should examine emotion regulation strategies, weight. Those results contribute to a growing body of including rumination, and explore whether alternative, more research indicating that loneliness affects health (Hawkley adaptive strategies for dealing with loneliness might not and Capitanio 2015), and are consistent with findings that lead to weight gain. Such work is important for informing higher loneliness leads to increased food consumption interventions focused on reducing loneliness and/or obesity. among female older adolescents (Rotenberg and Flood Future research will also want to explore how the phy- 1999). While we have not examined mechanisms linking siological effects of loneliness could influence eating loneliness and weight status, we have provided the first behavior including increasing the propensity to binge eat evidence that loneliness is directly related to increasing palatable foods and reducing one’s ability to track food weight for females upon entry into early adolescence. consumption. Such future work should take into account the The finding that loneliness reduces weight for male moderating effects of sleep and physical activity as those adolescents might suggest that loneliness serves as an factors affect weight metabolism and have been shown to be inhibitor of food consumption among those male adoles- deficient in lonely adolescents (Harris et al. 2013; Pels and cents; in contrast, loneliness may have disinhibited food Kleinert 2016b). The impact of our findings for interven- consumption for the female adolescents, a significant pro- tions to prevent obesity is clear – targeting early adoles- blem for those girls already overweight or obese females. cents, particularly females, who are lonely, could help in the The reduction in weight by male adolescents with high fight against obesity. weight status in the current sample may be demonstrable of a heightened awareness that they had to lose weight to Stability of Loneliness and Weight Status Over Time improve their social connections or were simply more motivated to do so. But, it may be the case that boys in the Results also support the stability of weight status and lone- current sample simply dealt with the stigmatization of liness over time. The relative stability of the constructs over overweight by inhibiting eating as a way to copy with time supports previous research (Pryor et al. 2011;van negative emotions. It is also possible that boys do not Dulmen and Goossens 2013). The strong stability of weight experience psychosocial effects until norms around their status and loneliness reflects the difficult task that interven- ideal male body shape (that of muscle; Field et al. 2014) tion teams face in order to affect changes in those areas of materialize during puberty. Future research should examine health. After age 13 years, the stability of weight status is those effects. likely to change due to physiological changes that accom- The weight gains we saw in girls are likely to be the pany puberty, and future work will want to examine how the outcome of engagement in emotional eating, a maladaptive onset of puberty impacts the prospective associations method of alleviating negative emotions (Haedt-Matt and between weight status and loneliness, exploring the impact Keel 2011), that they used to cope with feelings of lone- for male and female adolescents separately. Given that for liness. Goldschmidt et al. (2017) found that poor emotional females, advanced pubertal maturation is associated with awareness and limited access to adaptive emotion regulation internalizing problems, explained exclusively in terms of strategies contribute to emotional eating in adolescent environmental influences (Marceau et al. 2012), we might females, so further research will want to establish whether expect that the prospective association between weight status emotional eating and poor emotion regulation can help and loneliness is also moderated by pubertal timing for girls. 1394 Journal of Youth and Adolescence (2018) 47:1385–1397 The time interval between data collection waves in the study introduced confounds into the study. Equal time intervals varied from one to two years, but the effects were seen in the between measurement points would be preferable (Van- one-year time interval between age 12 to age 13 (T2 to T3). derWeele et al. 2011). Missing cases were dealt with That age could be a key sensitive period for peer relationship through Listwise deletion because the Data Missing Com- problems to impact weight status, future longitudinal studies pletely At Random (MCAR) tests proved significant. Our may consider one-year time intervals between waves to give chosen method is a less favored method of data cleansing a more nuanced picture of magnitude and persistence of and could have introduced bias in the sample by altering the effects over time (VanderWeele et al. 2011). standard error estimates for sub samples due to non-random data (Allison 2002). However, the use of bootstrapping Income Sufficiency and Effects on Weight Status and helps provide an assurance that the results are not spurious. Loneliness Areas for Further study In the current study, we explored the impact of income sufficiency on weight status and loneliness. We did not find Throughout this discussion we have highlighted important any effects linking recent social disadvantage, measured areas of future work. We noted the need to examine whether here in terms of income sufficiency, to high weight status. weight status exerts effects on loneliness only under specific That finding is consistent with other research (Lee et al. circumstances, including situations where there is a high- 2014) that found only poverty exposure prior to 2 years of level weight-based victimization or internalization of age had a robust association with adolescent obesity. appearance ideals. We also noted the need to further explore To our knowledge, this is the first study to examine the the gendered responses to loneliness, and how any differ- longitudinal associations between income sufficiency and ences between male and female adolescents are related to youth loneliness. We thought it might be the case that future gains or reductions in weight. children whose parents had limited income did not have In addition, in future studies, several further variables adequate time to spend with their children or have the could be controlled, including activity levels and onset of resources to support peer engagement activities, and chil- puberty. Ethnicity could not be explored as a moderator in dren from those families would experience increasing dis- the current study because the sample was not ethnically tance from parents and peers, and, thus, report loneliness. diverse. In future work, ethnicity should also be explored But, we did not find that adolescents whose families had given evidence that girls with high weight status in certain insufficient incomes as defined by Statistics Canada, ethnic groups suffer less from the negative effects of weight reported higher rates of loneliness. Further empirical work stigma than other groups (Mustillo et al. 2013). should examine income sufficiency in relation to loneliness among youth, examining whether that effect is found for children whose focus is more on parents as the main source Conclusion of support (Csikszentmihalyi and Larson 1984). The current study examined the concurrent and prospective Strengths and Limitations reciprocal relationships between weight status and loneliness, controlling for income sufficiency and gender. We found that This study has several strengths, including the large popu- both male and female adolescents with higher weight status lation sample, the robust anthropometric techniques used to reported increasing loneliness from ages 12 to 13 years, collect weight and height data, and the prospective nature of showing that society’s views of what constitutes an body the design. The current study followed children over a 3- ideal size and shape are communicated and reinforced by year period into early adolescence and allowed confirmation peers at this point in development. In addition, loneliness at of the temporal relationships between loneliness, weight age 12 years reduced weight for male adolescents from age status, and income sufficiency, and provided exploration of 12 to 13 years, and increased weight for female adolescents effects for male and female adolescents. The time period during that same period, suggesting that loneliness may serve provided a good test of the stability of the constructs during as an inhibitor of food consumption among male adolescents, a period of developmental transition. The simplicity, long- but may disinhibit food consumption for female adolescents, itudinal design, and youth sample in the current study add asignificant problem for already overweight or obese depth to our understanding of the links between weight females. Further work will want to explore the prospective status and social problems faced by young people during a effects further to determine whether the gendered effect is key period in development. specific to the current sample, and, if not, what that effect The study is not without limitations. The difference in tells us about gendered coping and peer group friendships interval lengths between data collection points could have among male and female youth. Journal of Youth and Adolescence (2018) 47:1385–1397 1395 The findings suggest early clinical attention to high weight change in a prospective community-based study of children fol- lowed up into adulthood. Archives of Pediatric Adolescent status and loneliness will be important and may have sig- Medicine, 160, 285–291. nificant effects for adolescent females. Reducing peer-related Bronfenbrenner, U. (2005). Making human beings human: Bioecolo- problems for those with high weight status will help reduce gical perspectives on human development. Thousand Oaks, CA: feelings of loneliness, but such problems reflect society’s Sage. Brylinsky, J. A., & Moore, J. C. (1994). The identification of body stigma attached to non-conformation to body ideals, making build stereotypes in young children. Journal of Research in such social norms hard to change. Thus, it will be more Personality, 28(2), 170–181. https://doi.org/10.1006/jrpe.1994. important to help adolescents develop resilient strategies of coping with peer criticism of non-ideal body size, reducing the Cacioppo, J. T., & Hawkley, L. C. (2003). Social isolation and health, with an emphasis on underlying mechanisms. Perspectives in chance that they will engage in emotional eating or experience Biology and Medicine, 46(3), S39–S52. https://doi.org/10.1353/ social stress, both of which would lead to weight gains. pbm.2003.0049. Cohen, J. (1988). Statistical power analysis for the behavioral sci- Acknowledgements This study uses data from the Quebec Long- ences (2nd ed.). Hillsdale, NJ: Erlbaum. itudinal Study of Child Development (QLSCD). Data collection for the Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied QLSCD was made possible through funding from the ministère de la multiple regression/correlation analysis for the behavioral sci- Santé et des Services Sociaux (MSSS) (Ministry of Health and Social ences. 3rd Edition New York, NY: Routledge. Services), the Lucie and André Chagnon Foundation, the ministère de Cole, T. J., Bellizzi, M. C., Flegal, K. M., & Dietz, W. H. (2000). la Famille (Ministry of Family), and the Institut de la statistique du Establishing a standard definition for child overweight and obe- Québec (the Institut). sity worldwide: International survey. British Medical Journal, 320, 1240–1243. Authors’ Contributions P.Q. conceived the study, performed statistical Cole, T. J., Flegal, K. M., Nicholls, D., & Jackson, A. A. (2007). Body analyses, interpreted the findings, drafted the manuscript, and made mass index cut offs to define thinness in children and adolescents: revisions to the manuscript based on reviewer feedback; R.H. and A. International survey. British Medical Journal, 335, 194 https:// M.E. helped to perform statistical analyses and interpret the findings, doi.org/10.1136/bmj.39238.399444.55. and contributed to the draft manuscript; J.A., M.B., and R.E.T. par- Csikszentmihalyi, M., & Larson, R. (1984). Being adolescent: Conflict ticipated in the study design, and edited the manuscript; M.B. and R.E. and growth in the teenage years. New York, NY: Basic Books. T. coordinated the original data collection. All authors read and Dallman, M. F. (2010). Stress-induced obesity and the emotional approved the final manuscript. nervous system. Trends in endocrinology and metabolics, 21(3), 159–165. https://doi.org/10.1016/j.tem.2009.10.004. Data Sharing Declaration The dataset analyzed during the current Denneel, S., Maes, M., Vanhalst, J., Bijttebier, P., & Goossens, L. study is not publicly available, but it is available from Institut de la (2018). Developmental change in loneliness and attitudes toward statistique du Québec (Information and Documentation Centre, Institut aloneness in adolescence. Journal of Youth and Adolescence, 47, de la statistique du Québec, 200, chemin Sainte-Foy, 3e étage Québec 148–161. https://doi.org/10.1007/s10964-017-0685-5. (Québec, G1R 5T4). de Onis, M., Onyango, A. W., Borghi, E., Siyam, A., Nishisa, C., & Siekmann, J. (2007). Development of a WHO growth reference for school-aged children and adolescents. Bulletin of the World Compliance with Ethical Standards Health Organization, 85, 660–667. Eschenbeck, H., Kohlmann, C. W., Dudey, S., & Schurholz, T. (2009). Conflict of Interest The authors declare that they have no conflict of Physician diagnosed obesity in German 6- to 14-year-olds. Pre- interest. valence and comorbidity of internalizing disorders, externalizing disorders, and sleep disorders. Obesity Facts, 2,67–73. Ethical Approval All procedures were in accordance with the ethical Field, A. E., Sonneville, K. R., & Crosby, R. D., et al. (2014). Pro- standards of institutional research committeesand with the 1964 Hel- spective associations of concerns about physique and the devel- sinki declaration and its later amendments or comparable ethical opment of obesity, binge drinking, and drug use among standards. This article does not contain any studies with animals. adolescent boys and young adult men. JAMA Pediatrics, 168, 34–39. https://doi.org/10.1001/jamapediatrics.2013.2915. Informed Consent Informed consent was obtained by the all indivi- Goldschmidt, A. B., Lavender, J. M., Hipwell, A. E., Stepp, S. D., & dual participants included in the study. Keenan, K. (2017). Emotion regulation and loss of control eating in community-based adolescents. Journal of Abnormal Child Open Access This article is distributed under the terms of the Creative Psychology, 45, 183–191. https://doi.org/10.1007/s10802-016- Commons Attribution 4.0 International License (http://crea 0152-x. tivecommons.org/licenses/by/4.0/), which permits unrestricted use, Goldschmidt, A. B., Sinton, M. M., Aspen, V. P., Tibbs, T. L., Stein, distribution, and reproduction in any medium, provided you give R. I., Saelens, B. E., & Wilfley, D. E (2010). Psychosocial and appropriate credit to the original author(s) and the source, provide a familial impairment among overweight youth with social pro- link to the Creative Commons license, and indicate if changes were blems. International Journal of Pediatric Obesity, 5, 428–435. made. https://doi.org/10.3109/17477160903540727. Grow, H. M. G., Cook, A. J., Arterbarn, D. E., Saelens, B. E., Drewnowski, A., & Lozano, B. E. (2010). Child obesity asso- References ciated with social disadvantage of children’s neighborhoods. Social Science and Medicine, 71, 584–591. https://doi.org/10. Allison, P. D. (2002). Missing data. Thousand Oaks, CA: Sage. 1016/j.socscimed.2010.04.018. Anderson, S. E., Cohen, P., Naumova, E. N., & Must, A. (2006). Hanlon, E. C., Tasali, E., Leproult, R., Stuhr, K. L., Doncheck, E., de Association of depression and anxiety disorders with weight Wit, H., & Van Cauter, E. (2016). Sleep restriction enhances the 1396 Journal of Youth and Adolescence (2018) 47:1385–1397 daily rhythm of circulating levels of endocannabinoid 2- and Adolescence, 41, 1111–1126. https://doi.org/10.1007/ arachidonoylglycerol. Sleep, 39, 653–664. https://doi.org/10. s10964-012-9762-y. 5665/sleep.5546. Markey, C. N. (2010). Invited commentary: Why body image is Haedt-Matt, A. A., & Keel, P. K. (2011). Revisiting the affect reg- important to adolescent development. Journal of Youth and ulation model of binge eating: A meta-analysis of studies using Adolescence, 39, 1387–1391. https://doi.org/10.1007/s10964- ecological momentary assessment. Psychological Bulletin, 137, 010-9510-0. 660–681. https://doi.org/10.1037/a0023660. Mooney, E., Farley, H., & Strugnell, C. (2009). A qualitative inves- Harris, R. A., Qualter, P., & Robinson, S. J. (2013). Loneliness tra- tigation into the opinions of adolescent females regarding their jectories from middle childhood to pre-adolescence: impact on body image concerns and dieting practices in the Republic of perceived health and sleep disturbance. Journal of Adolescence, Ireland (ROI). Appetite, 52, 485–491. https://doi.org/10.1016/j. 36, 1295–1304. https://doi.org/10.1016/j.adolescence.2012.12. appet.2008.12.012. 009. Moyer, A. E., Rodin, J., Grilo, C. M., Cummings, N., Larson, L. M., & Harrist, A. W., Swindle, T. M., Hubbs-Tait, L., Topham, G. L., Rebuffe-Scrive, M. (1994). Stress-induced cortisol response and Shriver, L. H., & Page, M. C. (2016). The social and emotional fat distribution in women. Obesity Research, 2, 255–262. https:// lives of overweight, obese, and severely obese children. Child doi.org/10.1002/j.1550-8528.1994.tb00055.x. Development, 87, 1564–1580. https://doi.org/10.1111/cdev. Mustillo, S. A., Budd, K., & Hendrix, K. (2013). Obesity, labeling, 12548. and psychological distress in late-childhood and adolescent black Hawkley, L. C., & Capitanio, J. P. (2015). Perceived social isolation, and white girls. Social Psychology Quarterly, 76, 268–289. evolutionary fitness and health outcomes: A lifespan approach. https://doi.org/10.1177/0190272513495883. Philosophical Transactions of the Royal Society B, 370, Nolen-Hoeksema, S., Wisco, B. E., & Lyubomirsky, S. (2008). 20140114 https://doi.org/10.1098/rstb.2014.0114. Rethinking rumination. Perspectives on Psychological Science, 3, Hayden-Wade, H. A., Stein, R. I., Ghaderi, A., Saelens, B. E., 400–424. https://doi.org/10.1111/j.1745-6924.2008.00088.x. Zabinski, M. F., & Wilfley, D. E. (2005). Prevalence, char- Pels, F., & Kleinert, J. (2016). Loneliness and physical activity: A acteristics, and correlates of teasing experiences among over- systematic review. International Review of Sport and Exercise weight children vs. non-overweight peers. Obesity, 13, Psychology. https://doi.org/10.1080/1750984X.2016.1177849. 1381–1392. Advance online publication. Heatherton, T. F., & Baumeister, R. F. (1991). Binge eating as escape Perlman, D., Peplau, L. A. (1981). Toward a social psychology of from self-awareness. Psychological Bulletin, 110,86–108. https:// loneliness. In S. Duck, R. Gilmour, (eds.) Personal relationships doi.org/10.1037//0033-2909.110.1.86. in disorder. (vol 3, pp 31–56). London: Academic Press. Hughes, M. E., Waite, L. J., Hawkley, L. C., & Cacioppo, J. T. (2004). Pryor, L. E., Tremblay, R. E., Boivin, M., Touchette, E., Dubois, L., A short scale for measuring loneliness in large surveys: Results Genolini, C., & Cote, S. M. (2011). Developmental trajectories of from two population-based studies. Research on Aging, 26, body mass index in early childhood and their risk factors: An 8- 655–672. year longitudinal study. Archives Pediatric and Adolescent Jones, D. C., & Crawford, J. K. (2006). The peer appearance culture Medicine, 165, 906–912. https://doi.org/10.1001/archpediatrics. during adolescence: Gender and body mass variations. Journal of 2011.153. Youth and Adolescence, 35, 257–269. https://doi.org/10.1007/ Puccio, F., Fuller-Tyszkieweicz, M., Ong, D., & Krug, I. (2016). A s10964-005-9006-5. systematic review and meta-analysis on the longitudinal rela- Juvonen, J., Lessard, L. M., Schacter, H. L., & Suchilt, L. (2017). tionship between eating pathology and depression. International Emotional implications of weight stigma across middle school: Journal of Eating Disorders, 49, 439–454. https://doi.org/10. The role of weight-based peer discrimination. Journal of Clinical 1002/eat.22506. Child and Adolescent Psychology, 46, 150–158. https://doi.org/ Puhl, R., & Brownell, K. D. (2001). Bias, discrimination, and obesity. 10.1080/15374416.2016.1188703. Obesity Research, 9, 788–805. https://doi.org/10.1038/oby.2001. Koenig, L. J., & Abrams, R. F. (1999). Adolescent loneliness and 108. adjustment: a focus on gender differences. In K. J. Rotenberg & Puhl, R. M., & Heuer, C. A. (2010). Obesity stigma: Important con- S. Hymel (Eds.), Loneliness in childhood and adolescence (pp. siderations for public health. American Journal of Public Health, 296–322). Cambridge: Cambridge University Press. 100(6), 1019–1028. https://doi.org/10.2105/AJPH.2009.159491. Lawler, M., & Nixon, E. (2011). Body dissatisfaction among adoles- Puhl, R. M., Latner, J. D., O’Brien, K., Luedicke, J., Forhan, M., & cent boys and girls: The effects if body mass, peer appearance Danielsdottir, S. (2016). Cross-national perspectives about culture and internalization of appearance ideals. Journal of Youth weight-based bullying in youth: Nature, extent and remedies. and Adolescence, 40,59–71. https://doi.org/10.1007/s10964-009- Pediatric Obesity, 11, 241–250. https://doi.org/10.1111/ijpo. 9500-2. 12051. Lee, H., Andrew, M., Gebremariam, A., Lumeng, J. C., & Lee, J. M. Puhl, R. M., & Luedicke, J. (2012). Weight-based victimization (2014). Longitudinal associations between poverty and obesity among adolescents in the school setting: Emotional reactions and from birth through adolescence. American Journal of Public coping behaviors. Journal of Youth and Adolescence, 41,27–40. Health, 104, e70–e76. https://doi.org/10.1007/s10964-011-9713-z. Little, R. J. A. (1988). A test of missing completely at random for Puhl, R. M., & Suh, Y. (2015). Stigma and eating and weight dis- multivariate data with missing values. Journal of American Sta- orders. Current Psychiatry Reports, 17,10 https://doi.org/10. tistics Association, 83, 1198e. 1007/s11920-015-0552-6. Lobstein, T., Baur, L., Uauy, R., & TaskForce, I. I. O. (2004). Obesity Qualter, P., Murphy, M., Abbott, J., Gardner, K. J., Japel, C., Boivin, in children and young people: A crisis in public health. Obesity M., Vitaro, F., & Tremblay, R. E. (2015). Developmental asso- Review, 5(Supplement 1), 4–104. https://doi.org/10.1111/j.1467- ciations between victimization and body mass index from 3 to 10 789X.2004.00133.x. years in a population sample. Aggressive Behavior, 41, 109–122. Marceau, K., Neiderhiser, J. M., Lichenstein, P., & Reiss, D. (2012). https://doi.org/10.1002/AB.21580. Genetic and environmental influences on the association between Qualter, P., Vanhalst, J., Harris, R., Van Roekel, E., Lodder, G., pubertal maturation and internalizing symptoms. Journal of Youth Bangee, M., & Verhagen, M. (2015). Loneliness across the life Journal of Youth and Adolescence (2018) 47:1385–1397 1397 span. Perspectives on Psychological Science, 10(2), 250–264. trajectories: An investigation of gene-community interplay over https://doi.org/10.1177/1745691615568999. early life course. Journal of Adolescent Health, 53, 328–334. Rankin, J., Matthews, L., Cobley, S., Han, A., Sanders, R., Wiltshire, https://doi.org/10.1016/j.jadohealth.2013.04.004. H. D., & Baker, J. S. (2016). Psychological consequences of Yang, K., & Victor, C. (2011). Age and Loneliness in 25 European childhood obesity: Psychiatric comorbidity and prevention. nations. Ageing & Society, 31,1–21. https://doi.org/10.1017/ Adolescent Health, Medicine, and Therapeutics, 7, 125–146. s0144686x1000139x. https://doi.org/10.2147/ahmt.s101631. Zawadzki, M. J., Graham, J. E., & Gerin, W. (2013). Rumination and Reitz, A. K., Zimmermann, J., Hutteman, R., Specht, J., & Neyer, F. J. anxiety mediate the effect of loneliness on depressed mood and (2014). How peers make a difference: The role of peer groups and sleep quality in college students. Health Psychology, 32, peer relationships in personality development. European Journal 212–222. https://doi.org/10.1037/a0029007. of Personality, 28, 279–288. https://doi.org/10.1002/per.1965. Rotenberg, K. J., Bharathi, C., Davies, H., & Finch, T. (2013). Bulimic symptoms and the social withdrawal syndrome. Eating Behavior, Pamela Qualter is Professor of Psychology for Education at 14, 281–284. https://doi.org/10.1016/j.eatbeh.2013.05.003. University of Manchester, Manchester Institute of Education, United Rotenberg, K. J., & Flood, D. (1999). Loneliness, dysphoria, dietary Kingdom. She leads a program of work focused on understanding the restraint, and eating behavior. International Journal of Eating predictors and outcomes of prolonged loneliness during childhood and Disorders, 25,55–64. adolescents. She is also interested in socio-emotional learning across Rotenberg, K. J., MacDonald, K. J., & King, E. V. (2004). The rela- development. tionship between loneliness and interpersonal trust during middle childhood. Journal of Genetic Psychology, 165, 233–249. Rotenberg, K. J., & Sangha, R. (2015). The relation between Bulimic Ruth Hurley is a PhD student at The School of Psychology, The symptoms and the social withdrawal syndrome during early University of Central Lancashire, United Kingdom. Her major adolescence. Eating Behaviors, 19, 177–180. https://doi.org/10. research interests include obesity, rumination, and emotional eating. 1016/j.eatbeh.2015.09.008. Sinton, M. M., Goldschmidt, A. B., Aspen, V., Theim, K. R., Stein, R. Alice M. Eccles is a PhD student at The School of Psychology, The I., Saelens, B. E., Epstein, L. H., & Wilfley, D. E. (2012). Psy- University of Central Lancashire, United Kingdom. Her major chosocial correlates of shape and weight concerns in overweight research interests include loneliness, peer relationships, and pre-adolescents. Journal of Youth and Adolescence, 41,67–75. adolescent health. https://doi.org/10.1007/s10964-011-9686-y. Turnbull, J. D., Heaslip, S., & McLeod, H. A. (2000). Pre-school children’s attitudes to fat and normal male and female stimulus Janice Abbott is a Professor of Psychology at The School of figures. International Journal of Obesity, 24, 1705–1706. https:// Psychology, The University of Central Lancashire, United Kingdom. doi.org/10.1038/sj.ijo.0801462. Her research interests are in the development and evaluation of patient- Vaillancourt, T. (2013). Do human females use indirect aggression as reported outcome measures (specifically, quality of life measurement). an intrasexual competition strategy? Philosophical Transactions of the Royal Society of London, 368(1631), 20130080 https://doi. Michel Boivin is Professor of Psychology at Université Laval, org/10.1098/rstb.2013.0080. Canada, Research Chair in Child Development and a Director of van Dulmen, M. H., & Goossens, L. (2013). Loneliness trajectories. Research Unit on Children’s Psychosocial Maladjustment (GRIP). He Journal of Adolescence, 36, 1247–1249. https://doi.org/10.1016/ leads a program of research on biological, psychological, and social j.adolescence.2013.08.001. components of child development, a program that is mainly anchored VanderWeele, T. J., Hawkley, L. C., Thisted, R. A., & Cacioppo, J. T. to population-based longitudinal studies of children, including the (2011). A marginal structural model analysis for loneliness: Quebec Longitudinal Study of Child Development and the Quebec Implications for intervention trials and clinical practice. Journal of Consulting and Clinical Psychology, 79(2), 225–235. https:// Newborn Twin Study. doi.org/10.1037/a002261. Vanhalst, J., Luyckx, K., Raes, F., & Goossens, L. (2012). Loneliness Richard E. Tremblay is Professor of Pediatrics, Psychiatry, and and depressive symptoms: The mediating and moderating role of Psychology at The University of Montreal and holds the Canada uncontrollable ruminative thoughts. The Journal of Psychology, Research Chair in Child Development. He is also a Director of the 146, 259–276. https://doi.org/10.1080/00223980.2011.555433. Research Unit on Psychosocial Maladjustment (GRIP), a Wang, J., Iannotti, R., & Luk, J. (2010). Bullying victimization among multidisciplinary research center. His research interests include child underweight and overweight U.S. youth: Differential associations development, education, prevention, family, parent-child relations, for boys and girls. Journal of Adolescent Health, 47,99–101. child psychopathology, aggressive behavior, delinquency, and https://doi.org/10.1016/j.jadohealth.2009.12.007. substance abuse. Wickrama, K. A. S., O’Neal, C. W., & Lee, T. K. (2013). Early community context, genes, and youth’s body mass index

Journal

Journal of Youth and AdolescenceSpringer Journals

Published: May 28, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off