Sleeve Gastrectomy for Youth With Cognitive Impairment or Developmental Disability

Participants and their guardians completed signed informed assent and consent, respectively, for the first 2 years of data collection. Afterward, signed informed consent was waived, and an information sheet was provided to families in which their ability to opt out of any research involvement was detailed. This was done with the permission of the institutional review board because the signed consent was the only document linking participant name to the database, ultimately increasing the risk for a confidentiality breach. No patients during the time period of data collection opted out of the research. Four families were not included in the research because the parents did not- speak English, and no translator was available at time of consent. The applicable institutional review board approved this research.

Adolescents were self-referred or referred for bariatric surgery by medical providers if they met medical criteria (a BMI ≥35 with a medical comorbidity or a BMI ≥40). Approximately 1 to 4 months before a potential surgery, the youth and their caregivers participated in a psychological evaluation. Clinicians referred to recommended guidelines in evaluating whether adolescents were well positioned to move forward with the surgery.^25,–27 This evaluation was conducted by a licensed clinical psychologist. Assessment of cognitive functioning via standardized intelligence testing was performed on a subset of patients between 2013 and 2016. From 2013 to 2015 the intelligence testing was performed for almost all patients attending preoperative visits and thereafter only for adolescents for whom there was a concern regarding cognitive ability or ability to assent or consent for surgery.

Standardized cognitive assessment was administered to 56 adolescents in the course of preoperative psychological evaluation between 2013 and 2016 out of a larger sample of 226 adolescents receiving bariatric surgery at 1 weight loss program from January 2010 to December 2017. Additionally, 8 patients were identified as having a previous diagnosis of intellectual or developmental disability and were included in this subset; these patients received surgery between 2010 and 2017. The adolescents included in the following analyses (N = 63) were primarily girls (65%) and African American (57%; 24% non-Hispanic white, 18% Hispanic, and 1% other race or biracial). The average age of participants was 17.7 years (SD = 2.0; range = 13–24 years). The pediatric patients in this study were affected by severe obesity, with an average BMI of 51.2 (SD = 8.6; range = 36–74). Approximately half of the participants were on Medicaid; no other indicators of socioeconomic status were recorded.

The data presented were gathered in the course of the preoperative psychological evaluation or during follow-up visits with the surgeon at 3, 6, 12, and 24 months postsurgery. Anthropomorphic data were gathered during these follow-up visits.

Age, race and/or ethnicity, and sex were extracted from the medical record. Participants reported these variables at their first medical visit.

Adolescents were identified as having CI/DD by 2 methods: Beginning in 2013 and through 2016, the Wechsler Abbreviated Scale of Intelligence, Second Edition (WASI-II) was added to the preoperative evaluation procedure to assess domains of cognitive functioning. This is a well-established measure of cognitive performance. It has been validated for use with individuals 6 to 90 years of age.²⁸ Adolescents with a full-scale IQ (FSIQ-4) <80 (n = 9), falling within the borderline to extremely low range, were included in the CI/DD group (see Table 1). Other adolescents had been identified in the medical record and by parent report in the preoperative psychological evaluation as having a diagnosed intellectual impairment or developmental disability (n = 8). Three of the participants presented with Down syndrome, and 5 presented with a previous mild-to-severe intellectual disability diagnosis of unknown etiology. For the current study, both the presence and absence of CI/DD and the FSIQ-4 score were used as predictors.

The percentage of excess BMI lost (%EBMIL) was calculated to indicate the degree of weight loss after surgery. This is a standard measure of weight loss after surgery and is calculated by assuming an ideal BMI of 25, subtracting this from preoperative BMI, and reaching the excess BMI.²⁹ At follow-up visits, the measured BMI is compared with the excess BMI to calculate the %EBMIL. We also examined weight loss using other metrics, including BMI postsurgery, change in BMI pre- and postsurgery, and percentage of total weight loss (%TWL). Because of loss to follow-up and the fact that some patients were not yet due for the 2 year follow-up, 94% of participants had weight loss data at 3 months postsurgery (n = 59), 63% at 6 months postsurgery (n = 40), 56% at 12 months postsurgery (n = 35), and 22% at 24 months postsurgery (n = 14). Table 1 provides a ratio of patients for whom data were collected compared with patients who were eligible for a clinic visit at each time point.

Descriptive statistics were calculated to characterize the sample, and t tests were conducted to evaluate between-group differences in baseline characteristics as well as differences in BMI, %EBMIL, change in BMI, and %TWL at each follow-up time point. Next, the growth trajectory of change in weight was examined by using nonlinear latent growth models (LGMs).^30,–33 In the LGM, outcome measures are %EBMIL, change in BMI, and %TWL measured at 3, 6, 12, and 24 months postsurgery. There are 2 latent growth factors in the model (ie, latent intercept growth factor [Eta0] and latent slope growth factor [Eta1]); for instance, the former represents the average level of %EBMIL at 3 months postsurgery, and the latter is used together with time scores to represent the rate of change in the %EBMIL between each time point. Whereas the time score was fixed to 0 and 1 at 3 and 6 months postsurgery for the purpose of model identification, the time scores at 12 and 24 months postsurvey were estimated. First, the unconditional LGM was estimated to explore the growth trajectory of %EBMIL (and, secondarily, to ensure that findings were consistent when models were estimated by using the other measures of weight outcome) between 3 and 24 months postsurgery. Two conditional LGMs were estimated, in which a dichotomous IQ measure (1: presence of intellectual disability, 0: absence intellectual disability) and the FSIQ-4 score, measured on a continuous scale, were used to predict the growth trajectory of %EBMIL (or other measure of weight outcome), respectively, controlling for covariates (eg, age, sex, ethnicity, and preoperative BMI).

The Bayesian approach was applied for model estimation because it has superior performance in small samples without reliance on asymptotic and data normality assumptions and is also more robust for handling missing data.^34,–36 The Bayesian approach is a full-information estimator in which all of the available data under a missing-at-random (MAR) assumption are used. Such a full-information approach is superior to the traditional approaches and similar response pattern imputation, especially in the context of longitudinal studies.^37,–39 Different from the missing completely at random assumption in the traditional statistical methods, MAR allows missingness to be related to observed covariates and/or outcome measures.^34,36,40 That is, any association between missing data and the observed covariates or observed outcomes does not violate the MAR assumption. The LGM was estimated by using Mplus 8.0.⁴¹ 

The goodness of fit of the model was assessed by the posterior predictive checking (PPC) method.⁴² If the model fits the data well, the 95% confidence interval (CI) of the difference between the observed and replicated χ² values should center around 0, and the posterior predictive P value (PPP) should be >.05.^35,42 Statistical inferences were made by examining the range of parameter estimates that captures 95% of the posterior probability distribution (ie, 95% Bayesian credibility interval). If the 95% CI of a parameter estimate does not cover 0, then the parameter estimate is statistically significant at the α = .05 level.^36,42 

Participants did not differ in covariate or outcome variables on the basis of their classification of having an intellectual disability or not (see Table 1). For all the LGMs, Bayesian estimation reached convergence after 2000 iterations, and all models fit the data well (PPC 95% CI covers 0, and PPP is close to .50). Selected results of the unconditional model are shown in Table 2. The means of the latent intercept and slope growth factors estimated from the model are Eta0 = 0.368 (95% CI: 0.326 to 0.410) and Eta1 = 0.104 (95% CI: 0.073 to 0.138), respectively, which means an average %EBMIL of 37% at 3 months postsurgery and increases of another 10% in 3 months to reach a mean change of 47% at 6 months. However, the change in %EBMIL slowed down at 6 months postsurgery such that the mean %EBMIL increased only 10% in 6 months and was 57% at 12 months. In the following year, there was almost no change in %EBMIL and the mean %EBMIL was 58% at 24 months postsurgery (see Fig 1).

The selected results of the 2 conditional LGMs are shown in Table 3. Presence of CI/DD did not have a significant effect on either the %EBMIL at 3 months postsurgery (β = .019; 95% Bayesian credibility interval: 0.059 to 0.095) or the rate of change in %EBMIL (β = .019; 95% Bayesian credibility interval: −0.046 to 0.082), controlling for covariates. There was a trend for a higher rate of change in %EBMIL for those individuals with CI/DD, suggesting that they may experience greater rates of weight loss over time than their typically developing peers. The preoperative BMI predicted the level of %EBMIL at 3 months postsurgery (β = −.01; 95% Bayesian credibility interval: −0.014 to −0.006) such that a lower baseline BMI predicted a greater %EBMIL over time. The second LGM, in which the FSIQ-4 score was used to predict the trajectory of %EBMIL, produced the same findings (see the right panel of Table 3). In other words, cognitive impairment, whether noted by the presence of CI/DD or measured with a standardized IQ measure at baseline, did not predict the %EBMIL after sleeve gastrectomy in adolescents. Latent growth modeling, as described above, was repeated by using BMI at each time point, change in BMI, and %TWL as outcome variables. Findings were consistent regardless of weight-outcome measure used.

In the current study, we examined the impact of developmental disability and impaired cognitive functioning on weight loss outcomes after LSG. Although individuals with CI/DD may more commonly experience obesity and associated comorbidities, our data suggest that borderline to extremely low IQ or previous diagnosis of developmental disability do not negatively impact weight loss after bariatric surgery in the short-term. In fact, the findings suggest no significant differences in overall weight loss (as defined by %EBMIL, BMI, change in BMI, or %TWL) or rate of weight loss and reveal a trend toward a better rate of weight loss in adolescents with CI/DD compared with their typically developing peers. Moreover, regarding general cognitive functioning, given that the FSIQ-4 score did not predict weight loss outcomes as anticipated, there seems to be no benefit to adolescents who are higher functioning regarding weight loss after surgery.

On the basis of these new data, LSG appears to be a viable and successful short-term weight management tool for adolescents with CI/DD, who are established as particularly vulnerable to obesity and secondary health concerns.^3,6 In fact, there may be advantages to undergoing surgery during adolescence rather than waiting until adulthood for this population. During adolescence, individuals with CI/DD are likely to have greater supports in place to assist them with tasks of daily living and may still be in the process of developing greater independence. It guarantees that the adolescent is not “going it alone.” This could translate to better cooperation with parental guidance regarding surgery requirements, including diet and exercise recommendations. Generally, there is also evidence that younger children respond better to behavioral interventions than adolescents do, suggesting that older youth may require a different treatment approach.⁴³ Bariatric surgery performed earlier in the trajectory of large weight gain has also been shown to lead to greater resolution of obesity, suggesting that waiting for adulthood can be detrimental.¹⁹ 

Concerns with offering bariatric surgery to individuals with CI/DD include the limited information available about the risks and benefits for these adolescents and questions about capacity to provide informed assent and/or consent. For the patients in the current study, we used the preoperative psychological evaluation to assess ability to provide assent and to ensure that a legal guardian was able to provide informed consent, and we discussed the ability to adhere to pre- and postoperative requirements, assessing the need for additional support (eg, additional education or pre- or postoperative inpatient stay). The multidisciplinary team considered this information and weighed it alongside other variables (eg, the success of previous weight loss efforts, patient BMI, medical comorbidities, etc) in offering guidance to families. Although recommendations regarding patient selection for bariatric surgery list cognitive disability as a potential reason to preclude surgery, it has been suggested that adolescents with CI/DD be considered on a case-by-case basis.^25,27 Given the paucity of published research in which outcomes of LSG in youth with CI/DD are documented, the current study adds valuable data to the discourse regarding this topic and reveals that the presence of a CI/DD alone should not be used to exclude patients from possibly having surgery.

One clear limitation of the current study is the small sample of adolescents with CI/DD who received surgery as well as fewer participants who provided data for later follow-up time points. Additionally, follow-up data only extended to 24 months postsurgery in the current study. It will be important in future research to evaluate progress across a longer time period after surgery. Although this provides less power for detecting longer-term impacts of surgery on the %EBMIL, the results immediately after surgery, and particularly the LSG procedure, are promising.

The current study reveals the importance of considering LSG as a treatment of severe obesity for all adolescents, regardless of intellectual ability. The preoperative assessment should consider risks and benefits of the surgery, ability to provide assent (and for a guardian, consent, if assent is unable to be obtained), and specific supports necessary for ensuring safety and the highest likelihood of success after surgery.

CI

confidence interval

CI/DD

cognitive impairment or developmental disability

Eta0

latent intercept growth factor

Eta1

latent slope growth factor

FSIQ-4

full-scale IQ

LGM

latent growth model

LSG

laparoscopic sleeve gastrectomy

MAR

missing-at-random

PPC

posterior predictive checking

PPP

posterior predictive P value

%EBMIL

percentage of excess BMI lost

%TWL

percentage of total weight loss