Technical Summary: Household Poverty and Physical Activity Endpoint Divergence Among U.S. Children and Adolescents

Problem Statement

While insufficient physical activity among children and adolescents is a well-documented global public health concern, existing research has largely focused on overall activity levels or the binary metric of meeting recommended guidelines. A critical gap remains in understanding the behavioral endpoints of physical activity distribution: complete inactivity (zero days) versus daily achievement of the recommended 60 minutes. Furthermore, the relationship between socioeconomic status and physical activity is often treated as linear, potentially obscuring complex patterns of behavioral heterogeneity within low-income populations. This study addresses whether household poverty is associated with a "polarization" of physical activity behaviors, where low-income children are disproportionately represented at both extremes of the activity spectrum (completely inactive vs. daily active), rather than simply exhibiting uniformly low activity levels.

Methodology

Data Source:
The study utilized cross-sectional data from the 2023–2024 National Survey of Children's Health (NSCH), a nationally representative survey of U.S. children. The analytic sample was restricted to children and adolescents aged 6–17 years.

Variables:

• Primary Exposure: Household poverty level, categorized by family income relative to the Federal Poverty Level (FPL): 0–99%, 100–199%, 200–399%, and 400%+ (reference). The study utilized multiply imputed poverty variables provided by the NSCH to handle missing data.
• Primary Outcomes:
  1. Zero-day physical activity: Reporting 0 days of at least 60 minutes of activity in the past week.
  2. Daily physical activity: Reporting at least 60 minutes of activity every day in the past week.
  3. Extreme physical activity: A composite variable representing either zero-day or daily activity.
• Covariates: Age, sex, race/ethnicity, caregiver education, family structure, health insurance coverage, neighborhood/school safety, access to parks/recreation centers, sports participation, and screen time (≥3 hours/day).

Statistical Analysis:
Analyses were conducted using StataMP 18.0, accounting for the complex survey design (weights, strata, primary sampling units).

1. Descriptive Analysis: Weighted means and percentages were calculated to characterize the sample across poverty levels.
2. Regression Modeling: Survey-weighted logistic regression models examined associations between poverty and the two primary endpoints (zero-day and daily activity).
3. Polarization Analysis: Multinomial logistic regression was used with the intermediate category (1–6 days) as the reference to estimate Relative Risk Ratios (RRRs) for falling into either endpoint.
4. Heterogeneity and Attenuation: Interaction terms between poverty and age groups (6–11 vs. 12–17 years) were tested. Sequentially adjusted models explored whether neighborhood conditions, sports participation, and screen time explained (attenuated) the poverty-inactivity association.

Key Results

1. Gradient Association with Inactivity:
Household poverty showed a graded association with zero-day physical activity. Compared to the 400%+ FPL group, the odds of reporting zero days of activity increased significantly as income decreased (OR = 1.70 for 0–99% FPL).

2. Non-Linear Association with Daily Activity:
Contrary to a simple "lower income equals lower activity" hypothesis, the lowest-income group (0–99% FPL) was also significantly more likely to report daily physical activity (OR = 1.22) compared to the highest-income group. Intermediate income groups did not show this elevated risk for daily activity.

3. Evidence of Socioeconomic Polarization:
Multinomial models confirmed that children in the 0–99% FPL group were significantly more likely to fall into both extreme categories (zero-day or daily) compared to the intermediate (1–6 days) category, relative to the highest-income group. This indicates a socioeconomic polarization where low-income children are concentrated at the behavioral endpoints rather than the middle.

4. Age Heterogeneity:
The polarization pattern varied by developmental stage. The association between poverty and complete inactivity was significantly stronger among children aged 6–11 years than among adolescents (12–17 years). The risk of complete inactivity in younger children was substantially attenuated after adjusting for sports participation and screen time, whereas a residual risk remained for the lowest-income adolescents.

5. Explanatory Factors:
Sequential modeling revealed that the lack of participation in sports teams/lessons and high screen time were the strongest predictors of zero-day physical activity. Adjusting for these factors significantly reduced the magnitude of the poverty-inactivity association, particularly for children aged 6–11.

Key Contributions

• Identification of Endpoint Divergence: The study moves beyond average activity levels to demonstrate that household poverty is associated with a bimodal distribution of behavior (polarization), where low-income children are overrepresented at both the "completely inactive" and "daily active" extremes.
• Developmental Nuance: It highlights that the mechanisms driving poverty-related inactivity differ by age, with childhood (6–11 years) being a critical period where inactivity is more strongly linked to modifiable factors like sports access and screen time.
• Methodological Rigor: The use of nationally representative data with complex survey weights and multiple imputation for poverty data ensures robust, generalizable estimates of these behavioral patterns.

Significance and Claims

The authors claim that these findings challenge the notion that low physical activity among low-income children is solely a result of limited health awareness or individual choice. Instead, the data suggests that poverty shapes physical activity through unstable and heterogeneous opportunities.

• Mechanism of Polarization: The "socioeconomic polarization" observed is interpreted as a reflection of resource instability. Some low-income children may be completely inactive due to barriers such as unsafe neighborhoods, lack of transportation, or inability to afford organized sports. Conversely, others may achieve daily activity through informal, unstructured, or low-cost means (e.g., active transport, neighborhood play) that do not require financial resources.
• Intervention Implications: The paper argues that reducing physical activity inequalities requires more than health education. It necessitates the creation of stable, safe, affordable, and organized opportunities for physical activity. Specifically, interventions should target the removal of practical barriers (cost, transportation, supervision) to sports participation, particularly for children aged 6–11, while addressing the complex interplay of peer support, autonomy, and environment for adolescents.

The authors explicitly state that due to the cross-sectional design, causal mediation cannot be established; however, the attenuation of poverty-related risks by sports participation and screen time points to these as critical, actionable targets for public health policy.