The Rise in Pediatric Obesity-Related Conditions and Costs in Public Insurance Programs: Evidence from Alabama

Abstract

Background:

The increase in pediatric obesity rates is well documented. The extent of corresponding increases in diagnoses of obesity-related conditions (Ob-Cs) and associated medical costs for children in public insurance programs is unknown.

Methods:

Retrospective claims data linked to enrollees' demographic data for Alabama's Children's Health Insurance Program (ALL Kids) 1999–2015 were used. Multivariate linear probability models were used to estimate the likelihood of having any Ob-C diagnoses. Two-part models for inpatient, outpatient, emergency department (ED), and overall costs were estimated.

Results:

The proportion of enrollees with Ob-C diagnoses almost doubled from 1.3% to 2.5%. The likelihood of diagnoses increased over time (0.0994 percentage points per year, p < 0.001). Statistically higher rates of increase were seen for minority and lowest-income enrollees and for those getting preventive well visits. Costs for those with Ob-Cs increased relative to those without over time, particularly inpatient and outpatient costs.

Conclusions:

Prevalence of Ob-C diagnoses and costs have increased substantially. This may partly be because of underdiagnoses/underreporting in the past. However, evidence suggests that underdiagnoses are still a major issue.

Introduction

The obesity epidemic in the United States has led to concerns about the rising rates of obesity-related conditions (Ob-Cs) among children and adolescents over the past decades, such as type 2 or adult onset diabetes, as well as hypertension, obstructive sleep apnea, and hyperlipidemia.^1–5

In 2000, a joint statement by the American Academy of Pediatrics and American Diabetes Association cautioned about the rising prevalence of type 2 diabetes among the pediatric population.⁶ Wang and Dietz⁷ reported a sharp increase in pediatric hospitalizations for obesity-related diseases between 1979 and 1999 and a more than threefold increase in obesity-associated annual hospital costs, from $35 million in 1979–1981 to $127 million in 1997–1999. More recently, two studies documented higher health service expenditures among children with obesity or overweight, versus normal weight, using the Medical Expenditure Panel Survey.^8,9 The growing prevalence of Ob-Cs and associated cost burdens are likely to have significant implications for public insurance programs that cover the pediatric population.

In 2017, a white paper released by Fair Health, Inc.,¹⁰ used insurance claims data from 2011 to 2015 from several private insurance companies to document changes in claims for Ob-Cs among pediatric enrollees. The paper reported a 109% increase in claims with diagnoses of type 2 diabetes among pediatric patients. They also documented a 110% increase in claims for prediabetes, 67% increase for hypertension, and 161% increase for obstructive sleep apnea. The authors concluded that the relatively recent increase in these claims, compared with the decades-long obesity epidemic, suggests that awareness and diligence among health care providers in documenting these diseases are relatively new phenomena. The study also highlighted incongruences between regions of the United States, with higher reported prevalence of pediatric obesity and claims for Ob-Cs. Alabama and Louisiana, in particular, ranked relatively low in such claims, even though these states have among the highest rates of obesity in the nation.^11,12 This suggests that in these states, the pediatric population may still may be subject to underdiagnosis and unmet needs.

In this study, we explore trends in the prevalence of diagnosed Ob-Cs among enrollees in the Alabama Children's Health Insurance Program (CHIP), ALL Kids, from 1999 to 2015. We examine whether the trends differ by race, family income level, and rural status, which would suggest a change in the likelihood of receiving a diagnosis for these children compared with their counterparts. We also explore whether the association between having a preventive well visit and receiving an obesity-related disease diagnosis has changed over time. Finally, we explore whether health care costs for enrollees with claims for Ob-Cs have increased relative to enrollees without evidence of these conditions. We are not aware of studies that have looked at the prevalence and changes in prevalence of diagnosed Ob-Cs among children in public insurance programs. By focusing on a Deep South state that has one of the highest rates of obesity in the nation, this study contributes valuable information on the implications of pediatric obesity for public insurance plans.

Methods

Originally, ALL Kids coverage was available to Alabama residents under age 19 with family incomes between 100% and 200% of the Federal Poverty Line (FPL). Beginning in October 2009, the income eligibility limits were increased to 300% of the FPL. In January 2014, the enrollee composition changed again as the Affordable Care Act required that all children up to 133% of the FPL be covered under Medicaid.

The ALL Kids program is administered by the Alabama Department of Public Health (ADPH), which contracts with Blue Cross and Blue Shield of Alabama (BCBSAL) for claims processing and management. Children enrolled in ALL Kids benefit from full medical, pharmaceutical, and dental coverage from the BCBSAL preferred provider network. Enrollees pay an annual premium and experience cost sharing in the form of copayments for selected services, although there are no upfront deductibles. Originally, children in families with incomes between 100% and 150% of the FPL (termed the low-fee group) face lower levels of cost sharing compared with children in families with incomes between 150% and 200% of the FPL (termed the fee group). Children in families with incomes between 200% and 300% of the FPL following the expansion in 2009 (termed the expansion group) have the same cost sharing as the fee group. The fourth group, comprising primarily Native American children (no fee), is federally exempted from all cost sharing. Beginning in January 2014, eligibility was adjusted based on modified adjusted gross income (MAGI)-based groups. This adjusted the minimum income level for eligibility upward and shifted the cost-sharing groups so that 146% to 156% represented the low-fee group, 157% to 208% represented the fee group, and 209% to 317% represented the expansion group.

We use claims data combined with enrollee demographic information from 1999–2015. Enrollee data are identified with study numbers in the database, and for purposes of data analysis, the unique identifiers are removed. The database is password protected and stored on a secure university server and encrypted. Only members of this research team responsible for empirical analyses have access to the data. The project was approved by the Institutional Review Board of the lead author's university.

We define Ob-Cs as having any one of the following six claims-based diagnoses: obesity, hypertension, prediabetes, diabetes, hyperlipidemia, and obstructive sleep apnea. The detailed International Classification of Diseases (ICD-10) codes used to identify each Ob-C are given in the Supplementary Appendix SA1. We present results on the prevalence of Ob-Cs separately for the full sample and also for the subsample of enrollees above 5 years of age (above-5) since diagnoses of obesity and Ob-Cs tend to increase as children grow older. Percentages with Ob-Cs for the subsample who have had at least one well-child visit in a given year are also shown since such visits may provide an opportunity to test for and diagnose such diseases. We also look at prevalence for the subsample with at least $100 in health care costs in a given year since diagnoses are only likely to occur when enrollees engage with the health care system.

We use multivariate regression analyses with linear probability models to identify predictors of the likelihood of an Ob-C. The models include enrollee characteristics as well as a linear time trend (measured in years). We examine whether the relationships between the likelihood of having an Ob-C and being a minority enrollee, an enrollee in the low-fee (or no fee) FPL group, and a nonurban enrollee have changed over time by interacting each of these variables in turn with the time trend. We also examine whether the relationship between having a well visit and Ob-C has changed over time using a similar interaction. All models control for other demographic characteristics such as gender and age. Linear probability models are preferred because coefficient estimates of the interacted variables in our model can be interpreted in a straightforward manner as effect sizes, whereas there are well-known challenges with deriving accurate effect sizes for interaction terms in logit and probit models.¹³ Nonetheless, we also run all models using logistic regression as a robustness check and present them in the Supplementary Appendix SA1.

Next, we use two-part models to estimate the association between having an Ob-C and health care costs and how that relationship has changed over time. This is an approach similar to that used by Trasande and Chatterjee.⁹ We do not limit our approach to costs specifically linked to an Ob-C claim since pediatric obesity is also linked to more use of other services such as mental health services.¹⁴ We run the models separately for outpatient, inpatient, ED, and total costs (inflation adjusted to 2015 dollars). The models include a binary indicator for Ob-C (1 if the enrollee has an Ob-C, 0 otherwise), a continuous time trend, and an interaction of the two. Models also control for other sociodemographic characteristics and well visits. The first part estimates the association of predictors with having any versus no health care costs in each service category and uses a logit model. The second part uses the same predictors and estimates their associations with the level of expense conditional on nonzero expense and uses a generalized linear model with a gamma distribution and log link.

The full models and estimation results are shown in the Supplementary Appendix SA1. Due to the complex nature of this model, the coefficients cannot be meaningfully interpreted as effect sizes. Hence, to allow readers to interpret results in terms of dollar figures, we present predicted outpatient, inpatient, ED, and total costs, with 95% confidence intervals, for enrollees with Ob-Cs (Ob-C = 1) and without Ob-Cs (Ob-C = 0) in 1999, 2007, and 2015. The statistical software Stata (v.15) is used for all analyses. The “Margins” command is used to obtain predicted values in the two-part models.

Results

Descriptive statistics in Table 1 indicate a steady increase in enrollees with Ob-Cs. For the full enrollee population, the prevalence increased from 1.2% to 2.9% in 2012 before declining to 2.4% in 2015. This decline coincides with migration of the lowest-income (<133%) ALL Kids enrollees to Medicaid under the Affordable Care Act's (ACA) mandated expansion for children. For the above-5 subsample of children, the corresponding increase is from 1.4% in 1999 to 3.3% in 2012, then declining to 2.8% in 2015. Very similar patterns are found when limiting the sample to those with at least $100 in health costs (results in Supplementary Appendix SA1). When the sample is limited to enrollees with a well-visit claim, rates go from 0.7% in 1999 to 3.1% in 2015 for all age groups and 0.7% to 4.7% for children 5 and older. Part of this sharp increase is likely driven by the changing characteristics of the subsample. For example, between 1999 and 2015, the percentage of enrollees with a well visit increased from 9.3% to 30.8% and the percentage of nonwhite enrollees with a well visit increased from 9% to 33.2%.

Table 1.

Percentage of ALL Kids Enrollees with Claims-Based Diagnoses of Any Obesity-Related Conditions, 1999–2015

	Percentage of enrollees with a diagnosis	Percentage of enrollees with a diagnosis, age 5+	Percentage of enrollees with a diagnosis (% of sample with well visit)	Percentage of enrollees with a diagnosis, age 5+ (% of sample with well visit)
1999	1.2	1.4	0.7 (9.27)	0.7 (6.46)
2000	1.2	1.4	0.6 (10.14)	0.9 (7.02)
2001	1.4	1.6	1.0 (11.64)	1.4 (8.19)
2002	1.5	1.7	1.0 (12.28)	1.8 (7.87)
2003	1.6	1.8	1.2 (12.52)	1.9 (8.16)
2004	1.7	2.0	1.4 (12.96)	2.2 (9.00)
2005	1.9	2.2	1.5 (13.83)	2.2 (9.42)
2006	2.2	2.5	2.1 (15.60)	3.0 (11.03)
2007	2.4	2.7	2.0 (17.64)	2.9 (12.79)
2008	2.5	2.9	2.5 (19.95)	3.4 (14.59)
2009	2.6	3.0	2.7 (20.33)	3.6 (15.23)
2010	2.7	3.1	2.8 (23.60)	3.8 (18.32)
2011	2.8	3.1	2.8 (25.08)	3.8 (19.33)
2012	2.9	3.3	3.1 (27.51)	4.2 (21.35)
2013	3.0	3.3	3.3 (27.50)	4.6 (21.47)
2014	2.5	2.9	3.1 (30.15)	4.7 (21.48)
2015	2.4	2.8	3.1 (30.76)	4.7 (21.79)

In an additional analysis not shown here, we find that the share of nonwhite enrollees among well-visit recipients increased from 34% in 1999 to 45% in 2015. Thus, the increased prevalence of Ob-Cs among this subgroup is likely the combination of a change in the type of enrollees receiving well visits, a change in physicians' propensity to diagnose or document Ob-Cs during well visits, and a change in the actual prevalence of these diseases.

Table 2 gives a breakdown of each Ob-C as well as the sample demographic characteristics for the pooled sample. Obesity diagnoses appear to be the most common (1.06% of sample), followed by hypertension diagnoses (0.60%) and diabetes diagnoses (0.41%). Approximately 50% of the sample is female, 41% are nonwhite, 54% belong to the low-fee group, and 35% are nonurban. A more detailed breakdown (not shown) indicated that enrollment in ALL Kids grew from 24,348 enrollees in 1999 to 73,295 in 2005 to 100,265 in 2012. After migration of the lowest-income enrollees to Medicaid, it declined to 65,580 in 2015. The proportion of nonwhite enrollees grew from 36% in 1999 to 43% by 2012 and remained at 42% in 2015.

Table 2.

Summary Statistics for Pooled Sample 1999–2015 for Type of Obesity-Related Condition and Sociodemographic Characteristics

	All enrollees	Enrollees age 5+ with $100 < in costs
	(N = 1,204,258)	(N = 995,993)
Obesity diagnosis	1.06%	1.23%
Diabetes diagnosis	0.41%	0.48%
Prediabetes diagnosis	0.006%	0.008%
Hypertension diagnosis	0.60%	0.70%
Hyperlipidemia diagnosis	0.28%	0.23%
Sleep apnea diagnosis	0.29%	0.27%
% Female	0.50	0.50
% Nonwhite	0.41	0.42
% Low-fee group enrollees	0.54	0.59
% Nonurban	0.35	0.35
Age (continuous variable)	10.74 (4.97)	12.36 (3.74)
% Has well visits	0.20	0.15

Table 3A–D shows results examining changes in the likelihood of Ob-Cs over time after controlling for other enrollee characteristics (gender, age and age-squared, race, FPL group, urban versus nonurban residence, and any well visit that year) and results examining how the changes over time vary by race, FPL category, nonurban status, and use of preventive care. The likelihood of an Ob-C increased 0.094 percentage points (p < 0.001) for each successive year for the full sample and a 0.102 percentage point increase (p < 0.001) was observed for the above-5 group (Table 3A).

Table 3.

Linear Probability Models Examining Time and Enrollee Characteristics as Predictors of a Diagnosis of an Obesity-Related Condition

	All enrollees		Enrollees age 5+
	N = 1,204,258		N = 995,993
	β	t-statistic	β	t-statistic
Specification 1: base model
Year	0.094	28.98	0.102	26.59
Age	0.119	10.21	−0.003	−0.10
Age²	0.006	11.68	0.011	9.37
Female	0.245	8.96	0.303	9.48
Nonwhite	0.510	18.14	0.570	17.32
Low-fee group	0.157	5.43	0.188	5.58
Nonurban	0.283	9.75	0.326	9.64
Well visits	0.952	26.18	1.295	28.10
Specification 2: minority status interactions
Year	0.080	19.51	0.086	17.63
Nonwhite^*year	0.036	5.67	0.040	5.38
Age	0.119	10.20	−0.003	−0.08
Age²	0.006	11.71	0.011	9.37
Female	0.245	8.97	0.304	9.50
Nonwhite	0.177	2.71	0.201	2.64
Low-fee group	0.159	5.50	0.190	5.65
Nonurban	0.283	9.77	0.326	9.65
Well visits	0.949	26.10	1.290	27.97
Specification 3: family income interactions
Year	0.069	15.56	0.076	13.45
Low-fee group^*year	0.050	7.82	0.047	6.17
Age	0.121	10.38	−0.000	−0.02
Age²	0.006	11.54	0.011	9.30
Female	0.245	8.97	0.304	9.49
Nonwhite	0.510	18.15	0.571	17.35
Low-fee group	−0.306	−4.64	−0.252	−3.19
Nonurban	0.281	9.68	0.324	9.58
Well visits	0.954	26.23	1.29	28.07
Specification 4: urban status interactions
Year	0.094	23.56	0.103	21.78
Nonurban^*year	0.000	0.13	−0.002	−0.29
Age	0.119	10.22	−0.003	−0.10
Age²	0.006	11.68	0.011	9.37
Female	0.245	8.96	0.303	9.48
Nonwhite	0.510	18.14	0.570	17.32
Low-fee group	0.157	5.42	0.188	5.58
Nonurban	0.275	4.18	0.346	4.51
Well visits	0.952	26.17	1.295	28.08
Specification 5: well-visit status interactions
Year	0.078	21.81	0.082	20.10
Well visits^*year	0.093	11.20	0.165	14.55
Age	0.118	10.08	−0.012	−0.35
Age²	0.007	11.85	0.012	9.64
Female	0.244	8.94	0.303	9.46
Nonwhite	0.509	18.08	0.568	17.29
Low-fee group	0.159	5.50	0.198	5.87
Nonurban	0.280	9.66	0.322	9.52
Well visits	0.006	0.06	−0.409	−3.25

Coefficient estimates multiplied by 100 to represent percentage point changes.

All variables that have estimates that are statistically significant at p < 0.05 are in bold.

There is a significant difference in the trend for nonwhites versus whites (Table 3B). The interaction term for nonwhite and year is significant both for the full sample (β = 0.036, p < 0.01) and the above-5 group (β = 0.040, p < 0.01). This implies that while for white enrollees, the likelihood of an Ob-C increased by an average of 0.080–0.086 percentage points each year, for nonwhite enrollees, it increased by additional 0.036–0.040 percentage points each year—that is, an ∼45% faster rate. The findings for the low-fee groups versus their counterparts are similar (Table 3C). For the nonlow-fee group, the likelihood of an Ob-C increases by 0.069 percentage points (p < 0.001) for the full sample and 0.076 percentage points for the above-5 group (p < 0.001), but for the low-fee group, it rises by additional 0.05 percentage points for the full sample and 0.047 percentage points for the above-5 group—about 70% faster rate. In contrast, there is no evidence that the trend increases at a significantly higher rate for nonurban versus urban enrollees (Table 3D). Finally, the interaction between having at least one well visit and the year variable is significant for the full sample (β = 0.093, p < 0.001) and for the above-5 group (β = 0.165, p < 0.001) and indicates a 125%–200% greater annual increase in the likelihood of an Ob-C for enrollees getting well visits versus those without well visits.

Logistic models computed for the full sample in the Supplementary Appendix SA1 support our findings with respect to direction and statistical significance of key predictor variables. Readers should be aware that the logistic model results presented are log odds, hence they are not comparable with coefficients from the linear probability models that depict actual effect sizes.

Estimated results for the full two-part model for inflation-adjusted health costs for outpatient, inpatient, ED, and total costs are shown in the Supplementary Appendix SA1. Results from the first part indicate that at the baseline, enrollees with an Ob-C are more likely to incur nonzero outpatient (β = 3.06, p < 0.001), inpatient (β = 1.36, p < 0.001), and ED (β = 0.55, p < 0.001) costs compared with their counterparts. However, while the likelihood of incurring any costs increases over time for all enrollees, there is no evidence that this likelihood increases at a statistically higher rate for those with Ob-Cs. In contrast, results from the second part of the model indicate that levels of costs (conditional on nonzero cost) increase over time for all enrollees, but they increase at a statistically higher rate for enrollees with an Ob-C; the interaction between Ob-C and year is statistically significant for outpatient (β = 0.11, p < 0.01), inpatient (β = 0.03, p < 0.01), and total (β = 0.02, p < 0.01) costs, although not for ED costs.

In Table 4, we show the adjusted predicted costs from the two-part model for 1999, 2007, and 2015. In 1999, those without Ob-Cs have average outpatient costs of $489.96, whereas those with Ob-Cs have costs of $1213.99—approximately 2.5 times higher. By 2015, those without Ob-Cs have average outpatient costs of $771.05, whereas those with Ob-Cs have costs of $2239.97—approximately 3 times higher. The increase is even larger for inpatient costs. In 1999, those with Ob-Cs had average costs ($1054.83) that were about four times higher than those without Ob-Cs ($269.09). By 2015, those with Ob-Cs had average costs ($2128.56) that were about seven times higher than those without Ob-Cs ($298.83). The ratio of total health care costs for those with Ob-Cs to those without Ob-Cs went from being about 2.5 times higher in 1999 ($2871.53 vs. $1161.83 on average) to 3.6 times higher in 2015 ($5670.61 vs. $1572.86).

Table 4.

Predicted Outpatient, Inpatient, Emergency Department , and Total Health Care Costs from Two-Part Models for Enrollees with and without an Obesity-Related Disease in Selected Years

	Outpatient costs (inclusive of copayments)
	Without obesity-related disease		With obesity-related disease
	Predicted value	95% CI	Predicted value	95% CI
1999	489.96	483.75–496.17	1213.99	1107.58–1320.41
2007	614.83	611.44–618.22	1649.03	1589.95–1708.12
2015	771.05	763.45–778.65	2239.97	2110.49–2369.45

	Inpatient costs (inclusive of copayments)
	Without obesity-related disease		With obesity-related disease
	Predicted value	95% CI	Predicted value	95% CI
1999	269.09	252.08–286.08	1054.83	823.49–1286.17
2007	284.21	276.15–292.28	1508.82	1358.36–1659.30
2015	298.83	281.64–316.02	2128.56	1765.55–2491.57

	ED costs (inclusive of copayments)
	Without obesity-related disease		With obesity-related disease
	Predicted value	95% CI	Predicted value	95% CI
1999	121.90	120.06–123.73	243.70	222.25–265.15
2007	136.47	135.54–137.40	297.66	286.75–308.57
2015	151.34	149.32–153.36	360.83	338.08–383.58

	Total costs (inclusive of copayments)
	Without obesity-related disease		With obesity-related disease
	Predicted value	95% CI	Predicted value	95% CI
1999	1161.83	1141.67–1181.99	2871.53	2522.63–3220.43
2007	1352.14	1341.80–1362.48	4036.23	3829.60–4242.85
2015	1572.86	1551.42–1596.30	5670.61	5212.50–6128.71

CI, confidence interval.

Discussion

In spite of widespread concerns about an epidemic of Ob-Cs in children and youth, there is limited information regarding the prevalence of such diseases in the pediatric population, trends over time, and implications for growth in health care costs. To our knowledge, no study has investigated trends in the prevalence of Ob-Cs or the cost implications for public insurance programs. We help fill that gap by analyzing changes in Ob-Cs and the changing association between health care costs and Ob-Cs in Alabama's CHIP, ALL Kids. Alabama has one of the highest reported rates of pediatric obesity in the nation, and prior research suggests that Ob-Cs might be relatively underdiagnosed in this state.

We find that pediatric Ob-Cs have grown sharply between 1999 and 2015 among ALL Kids enrollees. For example, we see anywhere from a doubling in the share of enrollees with such diagnosis (from 1.4% to 2.9%) for enrollees aged over 5 years to an almost sevenfold increase (from 0.7% to 4.7%) for enrollees aged over 5 years with at least one well visit. Furthermore, in multivariate regressions that control for other factors, we find a significant increase in the likelihood of Ob-Cs over time. We also find that the likelihood has increased more sharply for minorities and for low-fee group enrollees than their counterparts. Furthermore, there is a change in the association between well visits and an Ob-C, with children with a well visit being increasingly more likely to have a diagnosis in recent years than earlier years. Finally, we find that health care costs, particularly inpatient, outpatient, and total costs, have grown at substantially higher rates for enrollees with Ob-Cs than their counterparts.

We speculate that the increase in Ob-Cs is, at least in part, driven by underdiagnosing (or underreporting) of Ob-Cs in the early years, particularly among the more vulnerable ALL Kids enrollees. For example, based on trends computed from waves of the National Health and Nutrition Examination Survey (NHANES), obesity rates among 2–19-year-olds have increased from 13.5% to 19% between 1999–2000 and 2015–2016, which amounts to a 50% increase in the prevalence of obesity. Another study, using data from California, Colorado, Ohio, South Carolina, and Washington, found a 30% increase in prevalence of type 2 diabetes, from 0.34 per 1000 youth in 2001 to 0.46 per 1000 youth in 2009.¹⁴ While we do not expect Alabama to follow these exact trends, the divergence between the growth in prevalence of obesity and type 2 diabetes in these data and the ALL Kids data suggests that at least some of the increase in the incidence of Ob-Cs among ALL Kids enrollees is driven by improvements in diagnosis or more accurate reporting, rather than a change in the underlying prevalence of a condition.

Furthermore, a comparison of the percentage of enrollees with Ob-Cs in our study with other state-level data on pediatric obesity in Alabama suggests that there is still substantial underdiagnosis among the ALL Kids enrollees. For example, data from the women, infants, and children (WIC) participant and program characteristics survey suggest that 16.3% of 2–4-year-olds in Alabama have obesity, and data from the National Survey of Children's Health suggest that 18.2% of all 10–16-year-olds have obesity.¹¹ Underdiagnoses, or at least underreporting on medical records, by health care providers of obesity and Ob-Cs are well-recognized challenges. One review of two academic hospitals found that only about one-third of children with overweight actually received a diagnosis of overweight during outpatient visits.¹⁵ Another study using nationally representative data for 2005–2007 found that only 18% of 2–18-year-olds with obesity received an obesity diagnosis during preventive visits.¹⁶ Another study in Ohio found that 26% of children and adolescents with hypertension had that diagnosis documented in their electronic medical records.¹⁷

On a related note, data from NHANES III suggest that up to one-third of adults in the United States who have type 2 diabetes may be undiagnosed.¹⁸ A study published in 2002 found that pediatricians, pediatric nurse practitioners, and registered dietitians expressed concerns about Ob-Cs, but required additional training and resources to aid in weight management of children and adolescents.¹⁹ At the same time, it should be noted that ALL Kids has made notable progress in BMI assessment since the Children's Health Insurance Program Reauthorization Act of 2009 (CHIPRA) identified BMI as one of the core children's health care quality measures for state Medicaid and CHIPs—in 2013, ALL Kids was ranked in the top quartile of states in BMI assessment (52%–72% of enrollees).²⁰ This suggests that the issue of underdiagnoses or underreporting may become less of an issue in coming years.

The study has several limitations. Because we use claims data, we are unable to decipher the extent to which the changes over time in likelihood of Ob-Cs are driven by actual changes in the prevalence of the diseases, a change in the underlying characteristics of enrollees engaging with the health care system and getting preventive care, or a change in the propensity of physicians to test for, accurately diagnose, and report Ob-Cs. Furthermore, past research has found that Hispanic and Native American children are at particularly high risk of certain Ob-Cs, but we do not have a sufficiently large number of ethnic groups in our data to conduct separate analyses to see if the same patterns hold in Alabama. Finally, we use data from one state's stand-alone CHIP, and some caution must be exercised when generalizing these results. At the same time, we note that several other states have stand-alone CHIPs²¹ and comparable eligibility criteria,²² including Deep South states such as Georgia, Mississippi, and Texas, which also rank very high for pediatric obesity, and these findings may be especially pertinent for them.

In conclusion, results from ALL Kids suggest that the burden of pediatric Ob-Cs on public insurance programs has increased substantially, both in terms of the number of enrollees diagnosed with such conditions and the health care costs incurred by such enrollees. At the same time, there are likely still substantial underdiagnoses, although this may change with increasing awareness among providers and improved adherence to assessing BMI as part of the Child Core Set of quality measures for Medicaid and CHIP. This portends that even though the prevalence of pediatric obesity may be stabilizing—although that evidence is mixed^23–26—public insurance programs should anticipate seeing continued growth in Ob-C claims and costs associated with such conditions for the next several years.

Footnotes

Authors' Contributions

B.S., J.B., M.M., N.M., C.C., and D.B. conceived the concept and study design; P.S. and B.S. analyzed the data; and B.S., J.B., and D.B. interpreted the data. All authors were involved in writing the article and gave final approval of the submitted and published versions.

Acknowledgments

This project was conducted under contract with the Alabama Department of Public Health (ADPH Award number C80113037). ADPH approved the study design, provided claims data for analysis, and approved the report and its submission for publication.

Funding Information

This project was conducted under contract with the Alabama Department of Public Health (ADPH Award number C80113037).

Author Disclosure Statement

All authors received funding through the aforementioned ADPH contract, with the exception of C.C., who was Director of the Bureau of Children's Health Insurance at the Alabama Department of Public Health while this project was being conducted. The authors have no other conflicts of interest to disclose.

Supplementary Material

References

Sahoo

, Sahoo

, Choudhury

, et al. Childhood obesity: Causes and consequences. J Family Med Prim Care, 2015; 4:187–192.

Sorof

, Lai

, Turner

, et al. Overweight, ethnicity, and the prevalence of hypertension in school-aged children. Pediatrics, 2004; 113:475–482.

Andersen

, Holm

, Homoe

. Obstructive sleep apnea in obese children and adolescents, treatment methods and outcome of treatment—A systematic review. Int J Pediatr Otorhinolaryngol, 2016; 87:190–197.

Pulgarón

. Childhood obesity: A review of increased risk for physical and psychological comorbidities. Clin Ther, 2013; 35:A18–A32.

Kumar

, Kelly

. Review of childhood obesity: From epidemiology, etiology, and comorbidities to clinical assessment and treatment. Mayo Clin Proc, 2017; 92:251–265.

American Diabetes Association. Type 2 diabetes in children and adolescents. Pediatrics, 2000; 105:671–680.

Wang

, Dietz

. Economic burden of obesity in youths aged 6 to 17 years: 1979–1999. Pediatrics, 2002; 109:e81.

Finkelstein

, Trogdon

. Public health interventions for addressing childhood overweight: Analysis of the business case. Am J Public Health, 2008; 98:411–415.

Trasande

, Chatterjee

. The impact of obesity on health service utilization and costs in childhood. Obesity, 2009; 17:1749–1754.

10.

FAIR Health, Inc. (2017). Obesity and type 2 diabetes as documented in private claims data: Spotlight on this growing issue among the nation's youth. [online] Available at https://www.pedsendo.org/education_training/pediatric_endocrinologists/links/pdf/FAIR_Health_Obesity_and_Diabetes_White_Paper_Jan_2017.pdf Last accessed July 19, 2019.

11.

The State of Childhood Obesity. Robert Wood Johnson Foundation Web site. Available at https://www.stateofobesity.org/childhood/Published 2004. Updated 2019 Last accessed July 19, 2019.

12.

Childhood Overweight and Obesity Trends. National Congress of State Legislatures Web site. Available at www.ncsl.org/research/health/childhood-obesity-trends-state-rates.aspx Published 2019 Last accessed July 19, 2019.

13.

, Norton

. Interaction terms in logit and probit models. Econ Lett, 2003; 80:123–129.

14.

Estabrooks

, Shetterly

. The prevalence and health care use of overweight children in an integrated health care system. JAMA Pediatr, 2007; 161:222–227.

15.

Riley

, Bass

, Rosenthal

, Merriman

. Underdiagnosis of pediatric obesity and underscreening for fatty liver disease and metabolic syndrome by pediatricians and pediatric subspecialists. J Pediatr, 2005; 147:839–842.

16.

Patel

, Madsen

, Maselli

, et al. Underdiagnosis of pediatric obesity during outpatient preventive care visits. Acad Pediatr, 2010; 10:405–409.

17.

Hansen

, Gunn

, Kaelber

. Underdiagnosis of hypertension in children and adolescents. JAMA, 2007; 298:874–879.

18.

Diabetes in Children and Adolescents Work Group. An update on type 2 diabetes in youth from the National Diabetes Education Program. Pediatrics, 2004; 114:259–263.

19.

Story

, Neumark-Stzainer

, Sherwood

, et al. Management of child and adolescent obesity: Attitudes, barriers, skills, and training needs among health care professionals. Pediatrics, 2002; 110(Supplement 1):210–214.

20.

Centers for Medicare & Medicaid Services (2015). The State of Children's Health Care Quality in Medicaid and CHIP: Who Are the Higher-Performing States? [online] Available at https://www.medicaid.gov/medicaid/quality-of-care/downloads/child-core-set-hps-brief.pdf Last accessed July 19, 2019.

21.

CHIP State Program Information. Centers for Medicare & Medicaid Web site. Available at https://www.medicaid.gov/chip/state-program-information/index.html Last accessed July 19, 2019.

22.

Medicaid, Children's Health Insurance Program, & Basic Health Program Eligibility Levels. Centers for Medicare & Medicaid Web site. Available at https://www.medicaid.gov/medicaid/program-information/medicaid-and-chip-eligibility-levels/index.html Last accessed July 19, 2019.

23.

Ogden

, Carroll

, Lawman

, et al. Trends in obesity prevalence among children and adolescents in the United States, 1988–1994 through 2013–2014. JAMA, 2016; 315:2292–2299.

24.

Pan

, Freedman

, Sharma

, et al. Trends in obesity among participants aged 2–4 years in the Special Supplemental Nutrition Program for Women, Infants, and Children-United States, 2000–2014. MMWR Morb Mortal Wkly Rep, 2016; 65:1256–1260.

25.

Skinner

, Perrin

, Skelton

. Prevalence of obesity and severe obesity in US children, 1999–2014. Obesity (Silver Spring), 2016; 24:1116–1123.

26.

Skinner

, Ravanbakht

, Skelton

, et al. Prevalence of obesity and severe obesity in US children, 1999–2016. Pediatrics, 2018; 141:e20173459.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.04 MB