Income and Household Material Hardship in Children With Medical Complexity

We conducted a cross-sectional survey study of parents of CMC receiving primary care at a free-standing children’s hospital. The parents were participants of a research repository consisting of CMC enrolled in a specialized primary care clinic for CMC, as well as propensity score-matched CMC from the general primary care clinic. Of 363 repository participants, 249 (68.6%) parents completed the survey, of which 221 (88.8%) met the inclusion criteria by providing income data. Survey participants differed from nonparticipants by race and ethnicity and by primary care location (Supplemental Table 5). This study was approved by the hospital’s institutional review board.

Our main predictor was family SES, which was measured by using household income as a percentage of the federal poverty limit (FPL). Income was calculated as a percentage of the FPL based on household size. The study population was divided into 3 groups corresponding to the following categories: <50% FPL, 51% to 100% FPL, and >100% FPL. Survey items were taken from the National Survey of Children with Special Health Care Needs 2009–2010, as well as other validated survey instruments (Supplemental Table 4).^18,22–24 

Our main outcome was HMH measured in 3 areas: food, housing, and energy insecurity. Insecurity in any of these 3 domains was treated as HMH. The measurement of HMH using these domains has been previously validated by Children’s HealthWatch.^{17–19,24,25}  We also assessed HMH cumulatively (0–3 HMH),¹⁷  with 1 point for a positive response in each domain (Supplemental Table 4).

We obtained patient characteristics, including patient age, sex, insurance, and number of CCCs from the Pediatric Health Information System to reduce the number of items included in the survey. CCCs were identified by using Feudtner’s Complex Chronic Conditions Classification System.²⁶  We also collected child race and ethnicity, parental language spoken at home, and participation in the clinic for CMC within the general pediatrics clinic at the study site via a survey. We included the social construct of race and ethnicity because of its associations with SES and HMH.

We used descriptive statistics to calculate frequencies and the χ² test for bivariate comparisons. We performed adjusted logistic regression models to calculate the odds ratios and 95% confidence intervals (95% CI) of having ≥1 HMH. In the bivariate analyses, we included factors with P < .10 in the regression models. Analyses were conducted in SAS Enterprise Guide v8.3 (Cary, NC) at a significance level of P < .05.

Study population characteristics are demonstrated in Table 1. Of 221 included survey participants, the majority (60.2%) reported household incomes ≤100% of the FPL. Public insurance was more common among CMC with <50% FPL compared with families >100% FPL (90.7% vs 42.1%, P < .001). The percentage of patients who identified as non-Hispanic Black was also higher among families <50% FPL compared with families >100% FPL (45.4% vs 21.6%, P = .001).

A total of 40.9% of participants reported HMH, with energy insecurity being the most common HMH (28%; Table 2). Notably, 28.2% of families with income >100% FPL reported ≥1 HMH. In bivariate analyses, a higher percentage of patients with <50% FPL and 51% to 100% FPL had ≥1 HMH compared with patients with >100% FPL (<50% FPL: 50.0% vs >100% FPL: 28.2%, P = .01). Participants with income between 51% and 100% FPL reported a higher percentage of insecurity in food and energy compared with both higher and lower FPL groups; however, there was no significant difference in housing insecurity.

In adjusted analyses, patients from families with <50% FPL and 51% to 100% FPL had ∼75% higher odds of having ≥1 HMH compared with patients with >100% FPL (<50% FPL: odds ratio 1.74 [95% CI: 1.11–2.73], P = .02; 51% to 100% FPL: 1.73 [95% CI: 1.09–2.73], P = .02; Table 3).

In this study of 221 CMC receiving primary care at a tertiary children’s hospital, ∼41% of families had ≥1 HMH. Although households with incomes <100% FPL had 75% higher odds of having ≥1 HMH, 28.2% of families in the highest FPL group also had HMH. Although our findings show a correlation between FPL and HMH, income poverty led to the underestimation of the presence of household economic hardship. The fact that families with incomes 51% to 100% FPL reported a higher percentage of food and energy insecurity compared with both higher and lower FPL groups warrants further study. The proportion of families reporting HMH in our study is similar to or somewhat higher than that reported in large national surveys of the general pediatric population.^16,27  This may be due to the increased expenses experienced in families of CMC^28,29  or the overall low income of our study’s population.

This study provides insight into the complex balance of income and financial obligations in a uniquely vulnerable pediatric population.¹⁵  HMH was reported often, even affecting one-quarter of higher-income families of CMC, highlighting the importance of screening for HMH in CMC regardless of family SES. Our results indicate that FPL may allow us to predict the risk of HMH but generally leads to an underestimation of it. Specific sources of expenses, such as out-of-pocket health care costs, place a higher burden on families of CMC.^29,30  Additional health-related expenses, such as home or vehicle modification and respite care, may also be higher in this population.^28,31  Therefore, the authors of future studies should examine if HMH screening should include these sources of expenses. In addition, those authors should evaluate the impact of family HMH on the health and wellbeing of CMC and their families. Because HMH is concrete and intervenable, it would be important to understand if identifying HMH in families of CMC and obtaining resources to address them results in improvements in both HMH and clinical outcomes.

There are several limitations to note when considering our findings. Our study design was cross-sectional; therefore, we cannot make conclusions about causal relationships for any identified associations. Of surveyed participants, 11.2% did not provide income data and were excluded from our analyses; this may have biased our findings because there were some demographic differences between families who did and did not provide income data (Supplemental Table 5). Our sample was taken from only 1 children’s hospital, which may limit external generalizability. In addition, a majority of the study population was enrolled in a multidisciplinary CMC clinic. Although we controlled for this in our adjusted analyses, these patients would have received intensive support, which may have mitigated the influence of income on HMH. It would also reduce the generalizability of our findings to CMC populations without this intensive support.

Although poverty and HMH were correlated in this study, the use of income alone underestimated the presence of household economic hardship among families of CMC compared with HMH. The authors of future studies may use these findings to better measure the presence of HMH among CMC families. This measure may be useful in evaluating the effectiveness of interventions to mitigate the impact of SES on the health of CMC.