Accuracy of Weight Perception among Urban Early Adolescents with Uncontrolled Asthma and Their Caregivers

Abstract

Background

Obesity is associated with poor asthma outcomes; weight loss improves such outcomes. Inaccurate recognition of obesity may impede weight control.

Purpose

We examined perception of weight by early adolescents with uncontrolled asthma and their caregivers, and tested the relationship between medical visit frequency and accuracy of perceived weight status.

Methods

373 adolescents and their caregivers reported the adolescent’s height/weight and weight perception; caregivers reported healthcare utilization. We measured height/weight. Logistic regression modeled accuracy of weight perception.

Results

43.7% of the overweight/obese adolescents and caregivers accurately perceived weight status. BMI percentile (OR=1.19, CI=1.10–1.28) and total medical visits (OR=1.18, CI=1.05–1.33) were associated with higher accuracy in caregivers. Total medical visits (OR=0.84, CI=0.74–0.96) was associated with lower accuracy in adolescents.

Conclusions

Accurate perception of weight status was poor for overweight adolescents with uncontrolled asthma and their caregivers. Frequent medical visits were associated with improved caregivers’ but not adolescents’ perceptions.

Introduction

Childhood obesity is highly prevalent, particularly among ethnic minorities [1–3], and is co-morbid with asthma [4–6]. While this relationship is potentially bi-directional, most recently, it has been concluded that obesity usually precedes and worsens asthma [7, 8]. Overweight and obese youth with asthma have more frequent and more severe asthma symptoms [4, 9], are prescribed more asthma medications [4, 9, 10], have worsened lung functioning [4, 11, 12], have more Emergency Department (ED) visits and hospitalizations [9, 12], and have more school absences and days with activity limitations compared with non-obese peers [4, 13]. Weight reduction has been found to improve asthma severity and symptoms [12, 14]. Thus, an important component of asthma management is to focus on weight control.

A key tenet of the Health Belief Model [15–17] is that to improve health, individuals need to be aware of the severity of their conditions and acknowledge the benefits of taking steps to make positive health behavior changes. Because family-based behavioral pediatric obesity treatment programs are particularly efficacious [18], an important first step to engaging families in obesity programs is for both the early adolescent and caregiver to correctly perceive that the child is overweight or obese.

While accurate weight perception among adolescents has been found to be associated with weight loss behaviors [19] and reduction in weight [20], misperception of weight status or the discordance between people’s actual weight and their perception of their weight [21] serves as an impediment to weight loss [22, 23]. In fact, both children and their caregivers often under-assess the child’s weight status [21, 24–29]. Reasons for this are uncertain, but may be due to a person underestimating his/her personal risk, or an optimistic bias [30–32].

Factors associated with misperception of weight status in children and adolescents are understudied. Among adults, being male, non-Hispanic Black, of low-income status, or lacking a college degree have been associated with misperceiving weight status [33, 34]. In adolescents, racial/ethnic differences in perception of an ideal and/or healthy body weight may account for differences in weight perception [35–37]. In a study of over 4500 adolescents from Minnesota, African-American girls were less likely to perceive themselves as overweight than White girls, despite the higher prevalence of overweight and obese individuals in this population [38]. Moreover, in one study of over 1500 4^th and 7^th grade students in South Carolina, Black children selected significantly larger body sizes than White children for ideal adult and opposite gender body sizes [35]. Differences in weight perception may also have to do with community norms – children whose parents and schoolmates have higher BMI are more likely to misperceive their own weight [27, 39].

One potentially important, yet understudied, factor associated with weight status perception is frequency of medical visits. Given that height and weight are measured at the majority of these visits, each visit provides an opportunity for providers to identify and provide feedback to overweight and obese patients. While there is evidence that providers may fail to counsel patients adequately about their weight due to factors such as lack of time and training [40, 41], patients with frequent medical visits most likely receive some weight-related counseling. Data from the 2001 –2007 Medical Expenditure Panel Survey (n = 13,881) indicated that 47% of adolescent girls and 44% of adolescent boys were advised by a medical provider to eat healthy according to parent report. In addition, obese boys and girls were twice as likely to be advised to eat healthy. Pediatrician advice has been associated with better classification of child weight status by parents [42].

Even if providers fail to provide feedback and/or to counsel about weight, because anthropometric measurements are taken at most pediatric visits, having frequent medical visits provides more opportunities for the caregivers and early adolescents to learn and remember the child’s height and weight. Taking it a step further, becoming aware of the child’s actual weight may in turn prompt caregivers or the early adolescent to inquire about the significance of the weight (e.g., to ask how the child compares with other children or if it is a healthy weight). Thus, pediatric patients and their caregivers may be more likely to accurately perceive the child’s weight when the early adolescent is seen more frequently by health care providers.

Despite the potential importance of weight perception in promoting weight loss, as well as the importance of weight loss in improving asthma control, to our knowledge, no one has explored weight perception in early adolescents with asthma or in their caregivers. Consequently, this novel study sets out to determine the accuracy of weight perception in inner-city early adolescents with uncontrolled asthma and their caregivers, and the agreement of weight perception in these early adolescent-caregiver dyads. We also tested whether health care utilization is associated with accurate weight perception while controlling for potential confounding factors. We hypothesized that both early adolescents with uncontrolled asthma and their caregivers would have a high degree of inaccuracy in perceiving the early adolescents’ weight status, that there would be discordance among their opinions, and that more frequent visits to providers would be associated with more accurate perception.

Methods

Study Design and Participants

This study is a cross-sectional analysis of baseline data that originated from a larger, randomized controlled trial testing the efficacy of a school-based intervention to improve asthma management and asthma control [43]. All study procedures were approved by the Institutional Review Boards (IRBs) of the New York University School of Medicine, Columbia University, and New York City Department of Education and Department of Health and Mental Hygiene, and were conducted in accordance with the ethical standards of these IRBs, as well as with the Helinski Declaration of 1975, as revised in 2000. Prior to inclusion in the study, all participating caregivers gave their informed consent and provided permission for their child to participate, and all early adolescents provided written assent.

In the larger trial, we targeted Hispanic and African-American middle school students and their caregivers in New York City, recruiting families (n = 392) from 27 middle schools in Manhattan, Brooklyn, and the Bronx over four years (2005 – 2008); 12 schools (44%) had school-based health centers. To be eligible for the clinical trial, students had to be in the 6^th through 8^th grade, and their caregivers had to report that the early adolescent had a previous asthma diagnosis, had been prescribed asthma medication in the past 12 months, and had uncontrolled asthma. We used National Heart, Lung, and Blood Institute criteria [44] in place at the beginning of the study to determine whether students had uncontrolled asthma, which we defined as (1) daily daytime symptoms or night awakenings due to asthma symptoms three or more times a week, (2) daytime symptoms three to six days per week or night awakening three times per month to twice weekly, plus at least one urgent care visit to a doctor/clinic, hospital, or emergency department for asthma symptoms in the past year, or (3) less frequent, intermittent symptoms, plus two or more urgent care visits.

Study staff recorded baseline height and weight measurements of 379 early adolescents participating in the clinical trial. In this study, we excluded six students classified as underweight (i.e., Body Mass Index [BMI] < 5^th percentile) as they were not a large enough group to compare with the normal, overweight, and obese students, producing a sample of 373 early adolescent-caregiver dyads.

Procedures

Trained research assistants administered surveys to the early adolescents and caregivers at the student’s school either after-school or on weekends. The majority of the early adolescent-caregiver dyads completed surveys on the same day (88.4%; 330/373); those who did not, did so within a week. Early adolescents completed the survey in English; caregivers chose if they completed surveys in English or Spanish. English surveys were translated into Spanish and back translated by the research team which consisted of several native and non-native Spanish speakers who spoke different dialects of Spanish (e.g., Mexican, Puerto Rican); discrepancies were discussed until agreement on wording was achieved. Caregivers were compensated $30 for their baseline surveys, and early adolescents $10.

Research assistants measured the early adolescent’s height (to the nearest half inch) and weight (to the nearest half pound), without shoes or heavy articles of clothing. Staff used Homedics 315® digital scales to measure weight and a tape measure affixed to the wall to measure height. In some schools, the school nurse’s office had an analog scale and stadiometer which staff used for the measurements.

Study Instruments

Weight Classification

We calculated BMI using the Centers for Disease Control and Prevention (CDC) formula: (weight in pounds/[height in inches]²) x 703, standardizing BMI percentiles by age and gender [45]. We then classified the early adolescents into one of four BMI categories: (1) underweight (BMI < 5^th percentile); (2) normal (5^th ≤ BMI ≤ 85^th percentile); (3) overweight (85^th ≤ BMI≤ 95^th percentile); (4) and obese (BMI ≥ 95^th percentile). We further separated the obese group into BMI≥97^th percentile (herein referred to as “very obese”), as described in reports of national estimates using National Health And Nutrition Examination Survey (NHANES) data [1].

Weight Perception

Early adolescents and caregivers were asked, “Do you think that you/your child are/is overweight?” responding on a four-point Likert scale (0 = Not at all; 1 = Just a little; 2 = Pretty much; 3 = Very much). Because at the time of the study there were no validated measures assessing weight perception in early adolescents and different researchers asked this question in different ways [e.g., 19, 21–24, 27, 29, 46], this question was written by the investigators drawing on prior research. Two pediatric obesity researchers reviewed the question and response options for face validity.

After comparing the early adolescent’s and caregiver’s response to this item with the child’s actual BMI, we classified them as accurate or inaccurate. We defined accurate/inaccurate using three different classification systems based on face validity and prior literature [e.g., 19, 21–24, 27, 29, 46], and then ran regression models testing each classification system as part of a sensitivity analysis. For each classification system, the variables associated with accuracy of perception did not differ substantially. That is, the relationships for a given variable were in the same direction for each model, but they did not always reach statistical significance in some models.

We reviewed the definitions and results of these analyses with a committee consisting of a pediatrician, a behavioral medicine researcher, and a statistician, all of whom have experience in obesity research and/or treatment. There was agreement that: (1) the model using stricter accuracy criteria for the “very obese” had the most face validity (i.e., choosing “a little overweight” for this category did not count as accurate); (2) this same model was also the most innovative and clinically relevant given the recent attention to those who fall in this category and lack of literature on this particular sample; and (3) we should combine the overweight and obese early adolescents because the sample size of the obese group was relatively small once the very obese early adolescents were moved to their own group. Therefore, we defined accurate perception differently for overweight/obese and very obese early adolescents: for overweight/obese adolescents, an accurate response was defined as indicating that the adolescent was “just a little,” “pretty much” or “very much” overweight; for very obese adolescents an accurate response included that the adolescent was “pretty much” or “very much” overweight.

Participants were also asked to write what they thought they/their child weighed (in pounds). This number was compared to the staff-measured weight, and the difference between estimated and actual weight was calculated.

Asthma-related Medical Visits

Caregivers reported on the number of asthma-related urgent visits for the early adolescent’s asthma symptoms in the past two months, including acute medical visits to a doctor or clinic, emergency department visits, and hospitalizations. Additionally, caregivers reported the number of scheduled visits to the doctor’s office for routine asthma care. We summed the urgent and scheduled visits to compute the total number of asthma-related medical visits. While reliability statistics are not available for this variable, maternal report of health care use has been found to be moderately correlated to medical records, suggesting it is a valid way to assess health care use [47].

Asthma Symptom Days

The early adolescents were asked how many days in the past two weeks they had (1) experienced daytime asthma symptoms (wheezing, chest tightness, coughing, and shortness of breath), (2) woken at night due to asthma symptoms, and (3) been unable to participate in their usual activities due to their asthma symptoms. Following the procedures of other asthma investigators [48, 49], we computed a maximum number of asthma days score (range 0 – 14) by taking the highest of these three values. We used the “Maximum Asthma Days” score as an indicator of the early adolescents’ symptom control in the prior two weeks.

Demographics

The early adolescents reported their age, race/ethnicity, and gender. Caregivers answered questions concerning their relationship to the early adolescent (e.g., mother, father), their highest level of education completed, and employment status.

Statistical Analyses

We characterized accuracy of weight perception among early adolescents and their caregivers using descriptive statistics. We compared the proportion of early adolescents and caregivers who accurately perceived the child’s weight using a chi-square test. Early adolescent and caregiver dyad agreement regarding perception of weight was determined by calculating a weighted kappa (K_w) coefficient for the responses to, “Do you think you/your child are/is overweight?” Equal weighting within the kappa coefficient was employed in order to reflect the degree of difference between caregiver and early adolescent dyad responses [50]. We employed McNemar’s test of marginal homogeneity to determine if row and column marginal frequencies were equal [51]. A McNemar’s test compares the odds of early adolescent and caregiver dyads disagreeing in responses to, “Do you think you/your child are/is overweight?”

To test the relationship between health care utilization and accuracy of weight perception, using univariate logistic regression we modeled early adolescent and caregiver accuracy as functions of healthcare utilization, early adolescent demographics (e.g., age, gender, ethnicity, and BMI percentiles), asthma control, and early adolescent or caregiver estimated weight minus actual weight, respectively. We then constructed multivariate models including all variables listed above, even if not significant in the univariate models; we retained: (a) the demographic variables because they have been shown to predict accuracy in other studies (e.g., age, gender, ethnicity, BMI percentile) [27, 33, 34, 37, 52]; (b) maximum asthma days to ensure that significant findings were not a reflection of asthma control; and (c) estimated minus actual weight to control for the extent to which being able to accurately estimate weight is associated with accuracy in perception of overweight status. Early adolescent ethnicities were dichotomized as Hispanic versus all other ethnicities to aid in interpretation and avoid over-stratification of the data. Male was chosen as the reference category for all models involving early adolescent gender. Additionally, caregiver education was coded as to “less than high school” (the reference category), “high school/GED/vocational school” and “partial college/college graduate.” Statistical analyses were performed using the statistical computing software R-2.11.1 and SPSS. Statistical significance was judged at alpha = 0.05.

Results

Sample Characteristics

Table 1 reports early adolescent and caregiver characteristics. Most early adolescents self-identified as Hispanic (48%) or African-American/Black (34%); 46% were female and their average age was 12.8 years. Only 12% of caregivers graduated from college, and 51% were unemployed. Fifty-six percent of the early adolescents were overweight or obese, with 50% of overweight or obese early adolescents categorized as very obese (i.e., BMI percentile ≥97^th percentile); these rates are significantly higher than national estimates [13]. There were no significant demographic differences between normal weight and overweight or obese early adolescents with the exception of normal weight youth being slightly younger on average.

Table 1.

Early adolescent and caregiver characteristics by BMI weight classification at baseline. (N = 373)

	N (%)
	Normal Weight (N=164) (5th≤BMI<85th percentile)	Overweight/Obese (N=209) (BMI≥85th percentile)
Early Adolescent
Age*, mean ± SD, yr	12.9 ± 1.1	12.7 ± 1.0
Female	76 (46.3)	94 (45.0)
Ethnicity/Race
Hispanic	67 (40.9)	113 (54.1)
African-American/Black	64 (39.0)	62 (30.0)
Two or more ethnicities/races	14 (8.5)	19 (9.1)
Other	19 (11.6)	15 (7.2)
Caregiver
Relation to Early Adolescent
Biological or adoptive mother	147 (89.6)	187 (89.5)
Other	17 (10.4)	22 (10.5)
Highest level of education complete
Less than high school	53 (32.5)	72 (34.4)
High school/GED/vocational school	48 (29.4)	42 (20.1)
Partial college	42 (25.8)	70 (33.5)
College graduate	20 (12.3)	25 (12.0)
Employment status
Unemployed	92 (56.1)	100 (48.1)
Employed, part-time	18 (11.0)	24 (11.5)
Employed, full-time	54 (32.9)	84 (40.4)
Primary language
English	135 (82.3)	167 (79.9)
Spanish	29 (17.7)	42 (20.1)
Born in the US	96 (58.9)	123 (58.9)
Early Adolescent BMI Weight Classification
Overweight (≥85th percentile, but < 5thpercentile)		70 (18.8)
Obese (≥95th percentile, but < 97th percentile)		34 (9.1)
Very Obese (≥97th percentile)		105 (28.1)

^*Significant difference in normal weight and overweight/obese group means, p<0.05.

Accuracy of Weight Perception and Early Adolescent-Caregiver Agreement

Figure 1 shows the percentage of early adolescents and caregivers who accurately identified the early adolescent’s weight status by actual weight status; we included only families where both members of the dyad responded (n=267/373). Overall, while most of the early adolescents and caregivers accurately perceived the early adolescent’s weight status, with 73% (196/267) and 77% (206/267) being accurate, respectively, the dyads with normal weight adolescents comprised the majority of those who were accurate (54% [105/196] and 57% [118/206] of adolescents and caregivers, respectively). In fact, significantly smaller percentages of overweight, obese, and very obese early adolescents and their caregivers accurately identified the child’s weight status: 64% ([91/142] of overweight, obese and very obese early adolescents compared to 84% [105/125] of normal weight adolescents (χ² = 12.5, df = 1, p < 0.001), and 62% [88/142] of caregivers of overweight, obese and very obese early adolescents compared to 94% [118/125] of caregivers of normal weight early adolescents (χ² = 37.8, df = 1, p < 0.001).

Figure 1. Percentage of early adolescents and caregivers that accurately identified early adolescent’s weight status by BMI category. (N=267)*.

* Includes only dyads where both the early adolescent and caregiver responded.

Overweight, obese, and very obese early adolescents and their caregivers showed low to moderate agreement regarding the child’s weight status (K_w = 0.28, p < 0.001). Table 2 shows the percentages of correct and incorrect responses of early adolescent–caregiver dyads (overweight or obese youth only) when asked whether they thought the early adolescent was overweight or obese. Of note, only 43.7% of the early adolescents were perceived accurately by both the early adolescent and caregiver. For most dyads, either the early adolescent or the caregiver misperceived the early adolescent’s weight status (38.7%), or both the early adolescents and caregiver misidentified the early adolescent’s weight status (17.6%). When comparing the odds of an early adolescent being accurate while the caregiver was inaccurate to the odds of the opposite situation (i.e., caregiver was accurate and early adolescent was inaccurate), the McNemar’s test of marginal homogeneity indicated that both scenarios were equally likely (χ² = 1.09, df = 1 p = 0.30).

Table 2.

Agreement of weight perception accuracy between overweight and obese early adolescents and their caregivers. (N=142)

		Early Adolescent		Total
		Incorrect	Correct
Caregiver	Incorrect	25 (17.6%)	29 (20.4%)	54 (38.0%)
	Correct	26 (18.3%)	62 (43.7%)	88 (62.0%)
Total		51 (35.9%)	91 (64.1%)

Factors Associated with Caregiver and Early Adolescent Accuracy of Perception of Weight Status

Tables 3 and 4 show the distribution of independent variables we hypothesized were associated with early adolescent and caregiver accuracy of weight status perception, respectively. In the univariate models, ethnicity (OR (95% CI) = 2.11 (1.05, 4.23)) and total asthma-related medical visits (OR (95% CI) = 0.88 (0.78, 0.99)) were significantly associated with adolescent accuracy. However, after controlling for the effects of the other variables in the multivariate models, only total asthma-related medical visits remained significantly associated with lower odds of accuracy (see Table 3). For caregiver accuracy, total number of asthma-related medical visits (OR (95% CI) = 1.18 (1.05, 1.32)) and adolescent BMI percentile (OR (95% CI) = 1.18 (1.10, 1.27)) were associated with higher odds of accuracy in the univariate models, and these effects remained significant when controlling for all other variables in a multivariate model (see Table 4).

Table 3.

Model predictors of early adolescents’ weight perception accuracy.

	Inaccurate (N=51)	Accurate (N=91)	Adjusted OR (95% CI)
Early adolescent age, mean ± SD	12.6 ± 1.1	12.7 ± 1.1	0.99 (0.69, 1.42)
Early adolescent gender: female, N (%)	28 (51.0)	37 (40.7)	0.61 (0.28, 1.32)
Early adolescent ethnicity: Hispanic, N (%)*	22 (43.1)	56 (61.5)	2.07 (0.97, 4.44)
Caregiver Education:
High school/GED/vocational school VS. less than high school, N (%)	9 (17.6)	22 (24.2)	1.32 (0.43, 4.07)
Partial college/college graduate VS. less than high school, N (%)	27 (52.9)	35 (38.5)	0.50 (0.19, 1.29)
BMI Percentile, mean ± SD	95.5 ± 3.9	95.1 ± 4.2	0.98 (0.89, 1.08)
No. early adolescent reported maximum asthma days, mean ± SD	5.5 ± 3.9	6.2 ± 4.1	1.00 (0.91, 1.10)
Early adolescent estimated weight – actual weight, mean ± SD, lbs	−3.8 ± 10.1	−2.5 ± 6.6	1.02 (0.97, 1.07)
Total no. caregiver reported medical visits, mean ± SD (range)†,*	3.5 ± 3.5 (0 – 15)	2.4 ± 2.6 (0 – 11)	0.84 (0.74, 0.96)
Total no. caregiver reported urgent visits only, mean ± SD (range)¶, *,†	2.3 ± 3.0 (0 – 15)	1.3 ± 1.8 (0 – 9)	0.80 (0.68, 0.95)
Total no. caregiver reported scheduled visits only, mean ± SD (range)¶	1.3 ± 1.1 (0 – 4)	1.1 ± 1.3 (0 – 8)	0.84 (0.61, 1.15)

^*Significant difference in odds of being accurate from univariate logistic regression, p<0.05.

^¶Urgent visits and schedule visits were explored as separate model predictors of accuracy.

^†Significant difference in odds of being accurate from multivariate logistic regression, p<.05.

Table 4.

Model predictors of caregivers weight perception accuracy.

	Inaccurate (N=79)	Accurate (N=130)	Adjusted OR (95% CI)
Early adolescent age, mean ± SD	12.6± 1.0	12.7 ± 1.1	1.04 (0.76, 1.42)
Early adolescent gender: female, N (%)	36 (45.6)	58 (44.6)	1.12 (0.60, 2.09)
Early adolescent ethnicity: Hispanic, N (%)	41 (52.0)	72 (55.4)	1.07 (0.57, 2.01)
Caregiver Education:
High school/GED/vocational school VS. less than high school, N (%)	13 (16.5)	29 (22.3)	2.04 (0.82, 5.07)
Partial college/college graduate VS. less than high school, N (%)	36 (45.6)	59 (45.4)	1.59 (0.75, 3.36)
BMI Percentile, mean ± SD*,†	93.8 (4.6)	96.4 (3.3)	1.19 (1.10, 1.28)
No. early adolescent reported maximum asthma days, mean ± SD	6.3 ± 4.2	6.5 ± 4.1	1.02 (0.94, 1.10)
Caregiver estimated weight – actual weight, mean ± SD, lbs	−4.9 ± 7.9	−3.1 ± 8.1	1.04 (1.00, 1.08)
Total no. caregiver reported medical visits, mean ± SD (range)*,†	1.9 ± 2.4 (0 – 15)	3.2 ± 3.2 (0 – 15)	1.18 (1.05, 1.33)
Total no. caregiver reported urgent visits only, mean ± SD (range)¶, *, †	1.0 ± 2.2 (0 – 15)	1.9 ± 2.5 (0 – 12)	1.21 (1.04, 1.40)
Total no. caregiver reported scheduled visits only, mean ± SD (range)¶, *	0.9 ± 0.9 (0 – 8)	1.3 ± 1.2 (0 – 5)	1.38 (1.00, 1.90)

^*Significant difference in odds of being accurate from univariate logistic regression, p<0.05.

^¶Urgent visits and schedule visits were explored as separate predictors of accuracy.

^†Significant difference in odds of being accurate from multivariate logistic regression, p<.05.

To further understand the role of total asthma-related medical visits, we examined the distribution of urgent and scheduled asthma-related medical visits, the two types of visits that comprised the total number of asthma-related medical visits. Overall, caregivers of overweight, obese and very obese early adolescents reported fewer scheduled visits for asthma in the last two months with less variability (mean = 1.17, SD = 1.12, range = 0–8) compared to the number of asthma-related urgent visits (mean = 1.57, SD = 2.42, range = 0–15). We tested separate multivariate models using urgent visits only and scheduled visits only in place of total visits. When only urgent visits were used, results were consistent with models using total number of visits; scheduled visits did not predict early adolescent nor caregiver accuracy (see Tables 3 and 4).

Discussion

This study is the first to explore concordance between child-caregiver dyads regarding the perception of the child’s weight status and to examine factors associated with weight perception among urban early adolescents with uncontrolled asthma and their caregivers. With few exceptions, our hypotheses were supported. We found among the overweight and obese adolescents: (1) a high degree of misperception of the early adolescents’ weight status with only about 1/3 of both the early adolescents and their caregivers accurately perceiving overweight or obesity; (2) low to moderate agreement among the early adolescents and caregivers regarding the child’s weight status; (3) more frequent asthma-related medical visits was associated with more accurate weight perception for caregivers, but less accurate weight perception for the early adolescents; and (4) for caregivers only the early adolescents’ BMI percentile was associated more accurate weight perception.

The poor recognition of weight status we found among early adolescents with asthma and their caregivers is consistent with studies involving children without asthma where children and their caregivers often fail to accurately perceive the child as being overweight or obese [21, 24–29]. For example, in the National Longitudinal Study of Adolescents, a nationally representative, school-based prospective study of over 15,000 adolescents in grades 7–12 and their caregivers, 44% of teens that were obese by body mass index were not perceived as being obese by the adolescent nor the caregiver, and only 20% of obese teenagers were reported to be obese by both the adolescent and caregiver. An additional 6% were perceived as obese by the child only, and 30% by the parent only [24]. In a systematic review of parental perception of overweight status of children, Parry et al. found that in 19 of 23 studies, less than 50% of parents accurately perceived that their overweight child was overweight [28].

The inaccuracy in our sample is especially concerning because many were above the 97^th percentile for weight. Thus, they are at relatively higher risk for morbidity associated with both asthma and obesity, but potentially less likely to seek treatment for the obesity since only rarely did both the early adolescents and caregivers perceive the early adolescent as overweight.

While there has been no prior work examining discordance between adolescents and caregivers on identifying the child’s weight status, our finding of poor concordance among the early adolescents and caregivers regarding the child’s weight status is consistent with concordance research in other areas of health in child-caregiver dyads. For example, research has shown discordance among children and their caregivers regarding the frequency and severity of asthma symptoms [53–55] and pain [56–58], the presence or absence of headaches [59], and health-related quality of life [60]; the direction of this discordance is mixed with parents often under-reporting symptoms relative to their children, but often reporting more distress and worse quality of life.

One potential explanation for our findings regarding misperception of weight status is optimistic bias, or the propensity to underestimate one’s personal risk [30–32]. Several relevant mechanisms have been proposed to explain why unrealistic optimism may occur. One such mechanism is that optimistic biases are due to cognitive errors [31]. Individuals apply an incorrect norm [32], thereby lowering perceived risk. Our sample of early adolescents was predominately Hispanic and African-American, two groups where obesity is highly prevalent [1–3]. Because being overweight may be more the norm among these ethnic groups, the early adolescent and their caregivers may conclude incorrectly that they/their child is not at risk or not overweight. Support for this explanation comes from a study of over 3600 Canadian youth aged 9 – 16 where underestimation of weight status was associated having parents and schoolmates with higher BMI [27]. Similarly, a study of 4^th, 8^th, and 11^th grade children from Texas found that children were more likely to under-assess their own weight if there was a higher percentage of overweight children in their grade at school [39].

Unrealistic optimism also represents an attempt to protect oneself from harm [32]. If early adolescents or their caregivers acknowledge the child is overweight, and thus at more risk than others, their self-worth or competence is threatened. Therefore, they may cope by denying they/their child is overweight. Limited experience with risk also produces unrealistic optimism [31, 32]. This mechanism may explain our inconsistent finding regarding the early adolescent’s BMI being associated with greater accuracy by caregivers, but not the early adolescent. Caregivers may have more experience with obesity-related health risks, such as diabetes and cardiovascular disease, because they or family members they care for have the diseases. They understand the risk associated with increased BMI, and thus more accurately perceive their child’s weight. In contrast, the early adolescents have limited experiences with these diseases; these obesity-related problems have not occurred in them or their cohorts yet, so they feel they are exempt from this risk factor [30], and thus underestimate their risk – their perceived weight in this case. Related to this is the finding that unrealistic optimism is associated with egocentrism [31], a defining characteristic of adolescents but not adults [31, 61]. Egocentrism leads adolescents to believe that they, but not others, take steps to minimize risks and thus, are at less risk than others.

Our discrepant finding that more medical visits were associated with greater accuracy by caregivers, but less accuracy by the early adolescents, is difficult to explain given there are few relevant studies to guide interpretation. Thus, further studies are necessary to both confirm that it is reproducible and to explain reasons for this finding. The findings might be explained by different characteristics among early adolescents and adults that were not measured in this study (e.g., cognitive levels, attitudes toward chronic illnesses). Our observation that the majority of medial visits were for urgent, rather than scheduled, routine care, suggests that the nature of the visit may also be important, particularly if the early adolescents and caregivers have differential experiences. Urgent visits imply that the early adolescents are in distress, which may provoke anxiety in the early adolescent and/or the caregiver. This may impact how they each experience the visit. Therefore, studies that elucidate what happens during medical visits should be considered. Conducting exit interviews with early adolescents and caregivers after medical visits could tease out which potential factors (e.g., knowledge, attitudes, receiving weight-related feedback) contribute to the accuracy of weight perception. One might assess, for example, whether overweight and obese early adolescents and their caregivers believed they received feedback about the child’s weight during asthma-related medical visits, and how this compares to their medical providers’ perceptions.

Results need to be interpreted within the context of the study – a secondary data analysis of baseline data from a larger clinical trial. As such, it was not specifically designed to measure the impact of our examined covariates. Ideally, we would have liked to have ascertained whether the early adolescents and caregivers perceived the weight to be a health risk in general, and more specifically a risk for exacerbating asthma. Related to this is the fact that it was a cross-sectional study, and the causal nature of the relationship could not be established. It is feasible that more accurate perceptions of weight may lead to greater utilization of health care services, thus increasing medical visits. This concept merits further exploration.

Also, the sample was comprised primarily of Hispanic and African American urban early adolescents, thus limiting our ability to explore the potential differences by race/ethnicity, socioeconomic status, and developmental status (e.g., children versus early adolescents versus older adolescents). While we conducted exploratory analyses by race/ethnicity (i.e., Hispanic versus African American), we had a relatively small sample of overweight/obese Hispanic and African American early adolescents, and this precluded us from further subdividing by accuracy. This over-stratification yielded results that were difficult to interpret and were of questionable validity; therefore, we chose not to present them here. Similarly, we were unable to conduct analyses examining differences between different Hispanic ethnicities and levels of acculturation, as we did not collect this information. Such analyses are potentially important because differences among Hispanic ethnic groups have been found in asthma severity [62–64], as well as health care access [65, 66] and asthma-related health care utilization [64, 67]. Additionally, country of origin and acculturation have been found to impact behavioral health outcomes [68] and diet [69]. Therefore, future work should collect data on well-defined racial/ethnic groups and acculturation.

In summary, this study is the first to explore weight perception in a sample of urban early adolescents with co-morbid uncontrolled asthma and obesity. Our findings have important clinical implications. In order to first engage families in healthy weight reduction behaviors to help the early adolescent lose weight, both the adolescent and their caregivers must first accurately perceive the risk of being overweight. Interventions to help early adolescents and their caregivers first recognize this risk, especially as it relates to asthma, are warranted. Our findings also suggest that health care visits around asthma may provide important opportunities for medical providers to provide feedback to both the overweight pediatric patients and their caregivers about the early adolescents’ weight status and its potential impact on their asthma. This, in turn, may be an important first step towards motivating the family to help these patients to lose weight.