Association of Financial Distress With Nonmedical Out-of-Pocket Expenses in Those With Critical Illness

Abstract

OBJECTIVE:

Financial distress related to critical illness worsens clinical outcomes and is affected by objective financial difficulty or subjective perception of financial distress. Sources of financial distress include nonmedical out-of-pocket expenses (NOPEs; eg, food), but their impact on perceived financial distress is poorly understood, especially in pediatric critical care. We examine the association among overall NOPEs, their components, and perceived financial distress in families with children who have a critical illness.

METHODS:

This was a single-site cross-sectional survey (conducted from September 2021 to August2022) of caregivers of children admitted to the pediatric intensive care unit (PICU). Perceived financial condition was measured by the InCharge Financial Distress/Financial Well-Being tool and categorized as high (1.0–4.0), average (4.1–6.9), or low (≥7.0) distress. NOPEs included costs of transportation, food, childcare, and housing, reported as proportions of weekly income. We used ordinal logistic regression to examine the relationship between NOPEs and perceived financial distress.

RESULTS:

A total of 332 families were approached, and 279 provided consent (84%); 149 families returned surveys (53%). A total of 30% of families reported high distress, 45% reported average distress, and 25% reported low distress. Median proportion of total NOPEs relative to income was 34.7% for families with high distress, 12.7% for those with average distress, and 5.2% for those with low distress. The odds of increased financial distress was associated with the proportion of weekly income spent on total NOPEs (adjusted odds ratio [aOR], 1.02; 95% CI: 1.01–1.04) and on food (aOR, 1.22; 95% CI: 1.06–1.44).

CONCLUSIONS:

In this PICU population, proportions of weekly income spent on total NOPEs and, separately, on food were associated with perceived financial distress. Potential policy interventions could target reducing the burden of NOPEs to mitigate financial distress.

INTRODUCTION

Financial distress occurring in the context of illness negatively impacts well-being and health outcomes.^1–3 Illness-related financial distress is influenced by 2 components, ie, objective costs and subjective distress, together called “financial toxicity.”^4–7 Two contributors to objective costs are nonmedical and medical out-of-pocket (OOP) expenses. Nonmedical OOP expenses (NOPEs) include expenses such as transportation, housing, meals, and childcare, whereas copays and durable medical equipment are examples of medical OOP expenses.^8,9 Subjective distress, ie, the perception of financial distress, is associated with depression, decreased quality of life, and limited rehabilitation postdischarge.^2,3,10

Pediatric studies have demonstrated that families experience cost burden from both medical OOP expenses and NOPEs.^9,11–13 Perceived financial distress is also reported across the socioeconomic spectrum of caregivers of hospitalized children.¹⁴ However, there is little understanding of the distribution of financial distress and its risk factors specifically within the pediatric intensive care unit (PICU). Higher costs of PICU stays^15,16 could mean more financial vulnerability, making it imperative to understand the distribution of family financial distress and potential contributors. Moreover, the nature of critical illness itself may amplify this distress and vulnerability.

Our primary aim was to examine the role of NOPEs in the perceived financial distress of families with children who have a critical illness. We hypothesize that there is an association between the proportion of NOPEs relative to income and degree of perceived financial distress and that components of NOPEs may vary in their impact on perceived financial distress. Our secondary aim was to explore the relationship between markers of socioeconomic status (SES) and proportion of NOPEs to understand how external factors may affect NOPEs and perceived financial distress.

METHODS

Study Setting

We conducted a prospective cross-sectional study of family care-givers with children admitted to the PICU from September 2021 to August 2022 using electronic survey data collection. The study occurred at a tertiary care medical center in a large metropolitan area and focused on patients admitted to their 60-bed PICU. This study was approved by the institutional review board (2021–4759) of the children’s hospital.

A convenience sampling of family caregivers of children admitted to the PICU was performed. Caregivers were eligible if English or Spanish was their primary language, if they had an in-state residential address, and if they were recruited within 3 to 5 days of PICU admission. Using our institutional data, which showed an average length of stay of 3 days, we established this recruitment window to include families who were early in their admission and avoid those with brief PICU stays. Families were excluded if they lived at a long-term care facility or shelter because we planned to include neighborhood indicators of SES, as described later. We approached caregivers in-person or via telephone for verbal consent. Instead of written consent, caregivers who verbally consented received a disclosure form stating that completion of the survey indicated consent. To minimize recall bias, we asked caregivers to complete the survey before discharge from the PICU or hospital and followed-up in 4-day increments. No incentives were offered for study participation.

Data Collection

Study data were collected and managed using Research Electronic Data Capture (REDCap) tools hosted at the Northwestern University Clinical and Translational Science Institute. REDCap is a secure Web-based platform designed to support data capture for research studies.^17,18 At the time of verbal consent, caregivers provided an email address to which an individualized survey link was sent via REDCap. Additional demographic and clinical information was obtained from the child’s electronic health records.

Demographics

We used previously validated and published questions to create the survey.^9,12,13,19 The survey was professionally translated into Spanish. Survey content included self-reported demographics and financial status, financial predictor variables, and perceived financial distress using a validated tool²⁰ (Supplemental Figures 1 and 2). As a part of caregiver self-reported information, we asked social determinants of health (SDOH) screening questions regarding health care, utility, and housing insecurity.

Perceived Financial Distress

Our primary outcome was family financial experience, assessed using the 8-item InCharge Financial Distress/Financial Well-Being (IFDFW) questionnaire, a validated tool developed to measure distress related to financial condition.²⁰ This questionnaire is scored on a continuous scale (1–10) and, as recommended by the tool’s authors, we categorized participants as having high (1.0–4.0), average (4.1–6.9), or low/no (≥7.0) financial distress.²⁰

NOPEs

Our primary exposure was the proportion of caregiver-reported total NOPEs relative to self-reported income. Total NOPEs were calculated as the sum of costs associated with transportation, meals, childcare, housing, and miscellaneous items.^9,12,13 Families reported NOPEs for the 3 days before recruitment and selected their annual household income by ranges, as previously published.¹² To calculate a weekly income that matched the time period of reported expenses, we used the highest income from each category and divided that amount by 52 weeks. For the top-coded income category, we used a conservative approach, assigning an income of $150 000 to maintain similar income differences across all categories, as previously published.¹²

Clinical Variables

Clinical variables included the maximum Pediatric Logistic Organ Dysfunction (PELOD)-2 score during the first 72 hours of admission,^21,22 and presence of a chronic condition was determined using the complex chronic condition (CCC) classification system.²³ The PELOD-2 score is descriptive and assesses the severity of 5 organ dysfunctions at defined time points. The total score is on a scale of 0 to 33, with higher scores reflecting more organ dysfunction.^21,22 The CCC classification system is an International Classification of Diseases, Tenth Revision-based system used to identify conditions associated with life-long disability.²³ We used the presence of a CCC as an indicator of medical complexity.

Child Opportunity Level (COL)

In addition to clinical variables, we examined socioeconomic variables. We used the COL obtained from the Child Opportunity Index (COI) 2.0 based on geocoding the child’s zip code to the appropriate census tract using Maptitude 2016 (Caliper Corporation).²⁴ The COI is a composite 29-item measure of neighborhood-based opportunities grouped into the following 3 domains: education, health and environment, and social and economic. Race and ethnicity are not included in the COI. Each indicator is transformed into a z score, standardized, and weighted to reflect the strength of association between the indicator and outcome. The indicators are then combined to produce the aggregate COI score, available as metro-normed, state-normed, and nationally normed versions. The COI 2.0 also provides the COL, sorting census tracts into 5 ordered categories (ie, very high, high, moderate, low, and very low).²⁴ We used metro-normed COLs to better capture the landscape of this institution and grouped COL into 3 categories (ie, low/very low, moderate, and high/very high), as previously described.^25,26

Statistical Analysis

Statistical analyses were conducted using R Statistical Software version 4.1.2 (R Project for Statistical Computing). Descriptive statistics, including mean, SD, and count (percentage) were used to report demographics and clinical characteristics. We used χ² or Fisher exact tests (as appropriate) to assess differences in categorical variables and Kruskal-Wallis tests to assess differences in continuous variables among the financial distress groups (ie, high, average, and low).

We used ordinal logistic regression to model the association between the ordered, 3-category financial distress outcome and the following proportions of NOPEs relative to weekly income: total NOPEs, food, housing, transportation, childcare, and miscellaneous. The following variables were individually evaluated as possible confounders in subsequent regression models: caregiver age, insurance type (private, public, or other/none), caregiver education (high school or less or at least some college), marital status (single or partnered), 3-level metro COL, number of employed adults in the household, and self-reported race and ethnicity (ie, Asian, Black/African American, Hispanic/Latino, or white). We included race and ethnicity because historically minoritized populations may also be economically disadvantaged. We explored whether controlling for COL or race and ethnicity altered the relationship between proportion of NOPEs and perceived financial distress.

All adjusted ordinal logistic regression models included those variables significantly associated (P ≤ .05 threshold, to avoid overfitting) with both the primary outcome (ie, financial distress) and the primary exposure (ie, proportion of total NOPEs relative to income): caregiver age, insurance, marital status, and number of employed adults. We constructed 2 versions of the models: 1 including race and ethnicity, and 1 substituting COL for race and ethnicity. The Brant-Wald test was used to verify the proportional odds assumption of the ordinal logistic regression models.

We used linear regression to examine the association between the 3-category COL and proportions of NOPEs relative to weekly income. NOPE and medical OOP expense variables were log-transformed owing to skewed distributions. We tested for concordance between the IFDFW tool and SDOH questions to assess whether the IFDFW tool reliably identified families who are captured by these SDOH questions.¹⁹

Odds ratios (ORs) and 95% CI values are reported from the ordinal logistic regression models; β coefficients and 95% CI values are reported from the linear regression models. A two-tailed P value of less than or equal to .05 defined statistical significance in all analyses. A secondary analysis was performed to further examine families with multiple children.

RESULTS

Among 332 families who were approached, 279 (84%) provided consent, and 149 (53%) completed surveys. Demographic and clinical characteristics are described in Table 1. The mean caregiver age was 36.7 years, and the median reported upper bound of income was $59 999. Most families had higher than a high school education (77%), were privately insured (52%), and were partnered (72%). We considered race as a composite variable with ethnicity and observed the following: white (40%), Asian (22%), Black/African American (17%), and Hispanic/Latino (21%). The median total NOPEs and median proportion of weekly income spent on NOPEs for the whole cohort was $506 and 13%, respectively.

Table 1. Descriptive Characteristics.

Descriptive characteristics of the sampled population.

		Financial Distress
	Total Cohort (N = 149)*	High Distress (N= 44)	Average Distress (N= 67)	Low Distress (N= 38)	P-value^
Caregiver Age in Years, mean (SD)	36.7 (7.6)	34.0 (6.6)	37.8 (8.7)	37.9 (6.1)	p = 0.01
Household Income,n (%)
<$15,000	17 (11)	11 (25)	4 (6)	2 (5)
$15,000-$29,999	22 (15)	6 (14)	14 (21)	2 (5)
$30,000-$44,999	26 (18)	14 (32)	10 (15)	2 (5)
$45,000-$59,999	12 (8)	3 (7)	8 (12)	1 (3)
$60,000-$89,999	17 (11)	4 (9)	12 (18)	1 (3)
$90,000-$119,999	15 (10	4 (9)	8 (12)	3 (8)
≥ $120,000	39 (26)	2 (4)	10 (15)	27 (71)
Education,n (%)
Up Through High School	35 (23)	16 (36)	14 (21)	5 (13)	p = 0.04
Some college or more	114 (77)	28 (64)	53 (79)	33 (87)
Insurance,n (%)
Private	77 (52)	12 (27)	35 (52)	30 (79)	p <0.001
Public	59 (40)	27 (61)	26 (39)	6 (16)
No Insurance or Other	13 (9)	5 (11)	6 (9)	2 (5)
Marital Status,n (%)
Single	42 (28)	22 (50)	16 (24)	4 (11)	p <0.001
Partnered	106 (72)	22 (50)	50 (76)	34 (89)
Race/Ethnicity,n (%)
Asian	32 (22)	9 (21)	15 (22)	8 (21)	p <0.001
Black/African American	26 (17)	12 (27)	11 (16)	3 (8)
Hispanic/Latino	32 (22)	14 (32)	16 (24)	2 (5)
White	59 (40)	9 (21)	25 (37)	25 (66)
Metro Area Overall COL, n (%)
High	45 (35)	9 (22)	19 (34)	17 (52)	p = 0.09
Moderate	36 (28)	12 (30)	15 (27)	9 (27)
Low	48 (37)	19 (48)	22 (39)	7 (21)
Number (No.) of Household Adults	2 (1–8)	2 (1–8)	2 (1–6)	2 (2–6)	p = 0.35
No. of Household Children	2 (0–7)	2 (0–6)	2 (1–7)	2 (1–6)	p = 0.76
No. of employed adults in household	1 (0–4)	1 (0–3)	1 (0–4)	2 (0–3)	p = 0.003
Max 72-H PELOD score, mean (SD)	4.09 (2.78)	4.39 (2.76)	4.04 (2.80)	3.84 (2.79)	p = 0.60
CCC Flag Present,n (%)	124 (83)	36 (82)	59 (88)	29 (76)	p = 0.30
Families concerned for having their utilities discontinued,n (%)
Yes	29 (20)	19 (44)	7 (10)	3 (8)	p < 0.001
No	119 (80)	24 (56)	60 (90)	35 (92)
Families unable to afford health care,n (%)
Yes	26 (17)	14 (32)	9 (13)	3 (8)	p = 0.009
No	123 (83)	30 (68)	58 (87)	35 (92)
Families concerned about housing instability,n (%)
Yes	21 (14)	14 (32)	6 (9)	1 (3)	p < 0.001
No	128 (86)	30 (68)	61 (91)	37 (97)

^*Totals may not sum to 149 due to missing values and percentages may not sum to 100% due to rounding error.

^{^}Kruskal-Wallis rank sum test, Pearson’s Chi-squared test, Fisher’s exact test used as appropriate.

Financial Distress

Of the 149 families who completed surveys, 44 (30%) caregivers experienced high financial distress, 67 (45%) experienced average financial distress, and 38 (25%) experienced low/no financial distress (Table 1).

In the unadjusted univariable ordinal logistic regression models, odds of increased perceived financial distress were associated with having public insurance (OR, 4.95; 95% CI: 2.51–10.10), being single (OR, 4.27; 95% CI: 2.12–8.86), and having fewer employed adults in the household (OR, 0.50; 95% CI: 0.32–0.77). Race and ethnicity was associated with odds of increased financial distress as follows: Black/African American vs white (OR, 5.42; 95% CI: 2.19–13.90), Hispanic/Latino vs white (OR, 5.56; 95% CI: 2.41–13.30), and Asian vs white (OR, 2.30; 95% CI: 1.01–5.34; Table 2). Odds of increased financial distress were associated with the proportion of weekly income spent on total NOPEs (OR, 1.02; 95% CI: 1.01–1.04), food (OR, 1.27; 95% CI: 1.14–1.44), transportation (OR, 1.05; 95% CI: 1.01–1.09), childcare (OR, 1.02; 95% CI: 1.01–1.04), and miscellaneous items (OR, 1.06; 95% CI: 1.02–1.10).

Table 2. Unadjusted and Adjusted Multivariable Ordinal Logistic Regressions Using Race and Ethnicity.

Adjusted multivariable ordinal logistic regression using race and ethnicity. The first column is the unadjusted model. Each subsequent column (Model A through F) is a single model incorporating multiple variables.

	Unadjusted Model OR (95% CI)	Adjusted Models* aOR (95% CI)
		Model A	Model B	Model C	Model D	Model E	Model F
Proportion of NOPEs
Total NOPEs	1.02 (1.01–1.04)	1.02 (1.01–1.04)	-	-	-	-	-
Food	1.27 (1.14–1.44)	-	1.22 (1.06–1.44)	-	-	-	-
Housing	1.01 (0.99–1.03)	-	-	1.00 (0.99–1.03)	-	-	-
Transportation	1.05 (1.01–1.09)	-	-	-	1.04 (1.00–1.10)	-	-
Childcare	1.02 (1.01–1.04)	-	-	-	-	1.01 (1.00–1.03)	-
Miscellaneous	1.06 (1.02–1.10)	-	-	-	-	-	1.07 (1.02–1.13)
Age	0.96 (0.92–1.00)	1.00 (0.95–1.04)	1.00 (0.95–1.05)	0.99 (0.94–1.03)	0.99 (0.95–1.04)	0.99 (0.94–1.03)	0.99 (0.95–1.04)
Insurance
Private	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Public	4.95 (2.51–10.10)	1.90 (0.77–4.71)	1.64 (0.68–4.03)	2.12 (0.90–5.08)	1.76 (0.72–4.34)	2.03 (0.85–4.89)	2.24 (0.92–5.54)
Other or None	3.53 (1.15–11.20)	2.92 (0.73–12.01)	3.19 (0.83–12.83)	2.88 (0.74–11.66)	2.51 (0.63–10.28)	2.88 (0.73–11.71)	3.07 (0.78–12.44)
Marital Status
Partnered	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Single	4.27 (2.12–8.86)	1.72 (0.64–4.64)	1.58 (0.59–4.21)	1.97 (0.77–5.14)	1.98 (0.75–5.26)	2.17 (0.83–5.72)	1.80 (0.69–4.80)
Number of Employed Adults	0.50 (0.32–0.77)	0.71 (0.42–1.18)	0.67 (0.40–1.11)	0.62 (0.38–1.02)	0.68 (0.41–1.12)	0.64 (0.39–1.04)	0.67 (0.40–1.10)
Race/Ethnicity
White	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Black/African American	5.42 (2.19–13.90)	3.70 (1.27–11.15)	3.69 (1.27–11.08)	3.89 (1.38–11.30)	3.75 (1.32–11.05)	3.30 (1.16–9.66)	3.77 (1.32–11.16)
Hispanic/Latino	5.56 (2.41–13.30)	6.93 (2.57–19.67)	6.01 (2.28–16.58)	5.25 (2.03–14.08)	5.26 (2.02–14.22)	5.07 (1.95–13.67)	5.57 (2.12–15.27)
Asian	2.30 (1.01–5.34)	2.26 (0.92–5.65)	2.58 (1.08–6.30)	2.54 (1.07–6.16)	2.51 (1.06–6.08)	2.34 (0.98–5.70)	2.46 (1.02–6.05)

Boldface values indicate significance.

^*Each column (Model A through F) is a single model incorporating multiple variables. All models adjust for age, insurance, marital status, number of employed adults, and race/ethnicity.

Ordinal logistic regression models describing the associations of NOPEs with the 3 financial distress levels are presented adjusting for the following common covariates across all models: caregiver age, insurance, marital status, and number of employed adults in the household. Table 2 displays the adjusted model results adding race and ethnicity. Table 3 shows the adjusted model results substituting metro-level COL for race and ethnicity.

Table 3. Unadjusted and Adjusted Multivariable Ordinal Logistic Regressions Using Child Opportunity Level.

Adjusted multivariable ordinal logistic regression using the Child Opportunity Level. The first column is the unadjusted model. Each subsequent column (Model A through F) is a single model incorporating multiple variables.

	Unadjusted Model OR (95% CI)	Adjusted Models* aOR (95% CI)
		Model A	Model B	Model C	Model D	Model E	Model F
Proportion of NOPEs
Total NOPEs	1.02 (1.01–1.04)	1.02 (1.00–1.03)	-	-	-	-	-
Food	1.27 (1.14–1.44)	-	1.20 (1.03–1.43)	-	-	-	-
Housing	1.01 (0.99–1.03)	-	-	1.00 (0.98–1.02)	-	-	-
Transportation	1.05 (1.01–1.09)	-	-	-	1.03 (0.99–1.08)	-	-
Childcare	1.02 (1.01–1.04)	-	-	-	-	1.01 (1.00–1.03)	-
Miscellaneous	1.06 (1.02–1.10)	-	-	-	-	-	1.07 (1.02–1.14)
Age	0.96 (0.92–1.00)	0.98 (0.94–1.03)	0.99 (0.94–1.04)	0.97 (0.93–1.02)	0.98 (0.93–1.03)	0.98 (0.93–1.03)	0.98 (0.93–1.03)
Insurance
Private	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Public	4.95 (2.51–10.10)	2.18 (0.80–6.03)	1.75 (0.65–4.79)	2.29 (0.88–6.08)	2.14 (0.81–5.73)	2.19 (0.84–5.86)	2.28 (0.84–6.36)
Other or None	3.53 (1.15–11.20)	4.11 (1.13–15.95)	4.12 (1.16–15.78)	3.64 (1.02–13.96)	3.33 (0.92–12.88)	3.78 (1.05–14.51)	3.99 (1.10–15.48)
Marital Status
Partnered	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Single	4.27 (2.12–8.86)	1.46 (0.53–3.96)	1.49 (0.54–4.05)	1.82 (0.69–4.80)	1.82 (0.69–4.82)	1.89 (0.71–5.03)	1.42 (0.51–3.90)
Number of Employed Adults	0.50 (0.32–0.77)	0.67 (0.35–1.24)	0.63 (0.34–1.16)	0.62 (0.34–1.13)	0.68 (0.36–1.25)	0.66 (0.35–1.20)	0.65 (0.35–1.19)
Child Opportunity Level
Low/Very Low	1.50 (0.67–3.40)	1.61 (0.65–4.05)	1.64 (0.66–4.12)	1.47 (0.61–3.60)	1.37 (0.56–3.37)	1.45 (0.59–3.55)	1.62 (0.66–4.05)
Moderate	Ref	Ref	Ref	Ref	Ref	Ref	Ref
High/Very High	0.51 (0.22–1.17)	0.81 (0.32–2.08)	0.92 (0.36–2.33)	0.84 (0.33–2.12)	0.87 (0.35–2.18)	0.81 (0.32–2.04)	0.76 (0.29–1.97)

Boldface values indicate significance

^*Each column (Model A through F) is a single model incorporating multiple variables. All models adjust for age, insurance, marital status, number of employed adults, and Child Opportunity Level as a substitute for race/ethnicity.

In the models adjusted for race and ethnicity (Table 2), odds of increased financial distress remained associated with the proportion of weekly income spent on total NOPEs (adjusted OR [aOR], 1.02; 95% CI: 1.01–1.04), food (aOR, 1.22; 95% CI: 1.06–1.44), and miscellaneous items (aOR, 1.07; 95% CI: 1.02–1.13).

In the models adjusted for COL (Table 3), odds of increased financial distress remained associated with the proportion spent on food (aOR, 1.20; 95% CI: 1.03–1.43) and miscellaneous items (aOR, 1.07; 95% CI: 1.02–1.14). Costs and proportions of weekly income spent on NOPEs are presented in Table 4, and proportions of caregivers reporting NOPEs are presented in Supplemental Table 1.

Table 4. Median Total Dollar Amounts and Proportions of Weekly Income of Non-medical Out-of-Pocket Expenses.

Total dollar amounts and proportions of weekly income of NOPEs. Total dollar amounts of NOPEs are for the 3 days prior to recruitment.

	Total Cohort (N=149)	Financial Distress
		High Distress (N= 44)	Average Distress (N= 67)	Low Distress (N= 38)
Median NOPEs, $ (IQR)
Total NOPEs	293.00 (150.00,758.00)	761.00 (726.50,795.50)	240.00 (166.25,307.00)	442.00 (137.50,1728.00)
Food	30.00 (20.00,40.00)	30.00 (20.00,50.00)	25.00 (18.00, 32.50)	30.00 (16.25,48.75)
Housing*	0 (0, 259.25)	446.00 (368.00,463.00)	0 (0,0)	0 (0,167.00)
Transportation	50.00 (25.00,80.00)	60.00 (25.00,100.00)	50.00 (27.50,70.00)	52.50 (26.25,80.00)
Childcare*	0 (0,150.00)	45.00 (0,212.50)	0 (0,30.00)	0 (0,0)
Miscellaneous	20.00 (0, 100.00)	50.00 (0,100.00)	0 (0,65.00)	5.00 (0,100.00)
Median Proportion of Weekly Income, % (IQR)
Total NOPEs	13.00 (5.10,36.18)	34.67 (16.90,69.62)	12.71 (4.88,27.34)	5.20 (2.95,17.85)
Food	2.08 (1.08,3.76)	3.47 (2.17,6.93)	1.73 (0.95,3.47)	1.39 (0.91,2.08)
Housing*	0 (0,0)	0 (0,0)	0 (0,0)	0 (0,0)
Transportation	3.47 (1.39,7.80)	6.93 (3.32,11.70)	3.47 (1.73,7.28)	1.73 (0.48,3.38)
Childcare*	0 (0,8.67)	2.55 (0,28.60)	0 (0,0)	0 (0,0)
Miscellaneous	0.83 (0,6.93)	5.20 (0,11.56)	0 (0,5.49)	0.17 (0,4.12)

^*Childcare and housing expenses in the context of the total cohort. Not every family had additional children/childcare expenses or housing expenses

Of the families with multiple children, odds of increased financial distress were associated with being single (aOR, 7.29; 95% CI: 1.31– 46.21) and proportion of weekly income spent on total NOPEs (aOR, 1.02; 95% CI: 1.001–1.044; Supplemental Table 2).

COL

The 3-level metro COL was associated with the proportion of weekly income spent on transportation. Compared with families with moderate COL, the proportion of weekly income spent on transportation was 48.7% lower (95% CI: 2.8%–73.0%) for families with high COL (Supplemental Table 3).

SDOH

Testing for concordance between the IFDFW score and SDOH questions showed that, of families experiencing high financial distress, 44% reported concern regarding having their utilities discontinued, and 56% reported no such concern (P < .001). Of families experiencing low/no financial distress, 8% reported concern about having their utilities discontinued, and 92% reported no such concern (P < .001). The remainder of the results from this testing can be found in Table 1.

DISCUSSION

In this study of 149 caregivers of children with critical illness, the odds of increased financial distress were associated with the proportion of weekly income spent on total NOPEs, even after adjusting for covariates. Furthermore, we found that the proportion of weekly income spent on food remained associated with odds of increased financial distress independent of other risk factors. Although other NOPE categories had higher dollar costs (ie, housing), the models suggest that food expenses were the main driver of increased perceived financial distress. These additional costs are likely not the primary culprits but could amplify financial distress and could indicate opportunities for interventions. For already financially constrained families, absorbing additional health care expenses such as NOPEs may be untenable.²⁷

In our adjusted models, identifying as Black/African American or Hispanic/Latino was associated with having increased financial distress. Given that race and ethnicity are social constructs reflecting deep-rooted structural and systemic inequities, we used COLs in the second adjusted model because it is race-neutral. In those models, there was no association between COL and odds of increased financial distress. The differences in effect on financial distress demonstrate that financial distress may not be solely related to objective financial status but is instead a complex interplay of current situations and past experiences. Understanding these complexities requires future studies focused on individual experiences, environmental influences, and other factors to accurately capture the subgroup variations that exist.

By comparing the IFDFW tool against SDOH questions regarding utilities, health care, and housing, we demonstrated that perceived financial distress may be seen in families even if they do not report issues using common SDOH screening questions, revealing a subgroup with potentially unaddressed financial stressors that might be missed using only SDOH screenings. This may be because of differences between objective social disparities and subjective distress, highlighting that, although financial distress may be amplified in those with other social disparities, it does meaningfully exist across social strata. SDOH screenings may not identify all families experiencing financial distress; however, this distress influences outcomes.^2,3,10 Not accounting for perceived financial distress risks underrecognizing an important subgroup of vulnerable families. This is further supported by recent pediatric literature describing the high prevalence of debt and poor credit within families even before their child’s PICU admission.²⁸ This is key when considering the implications of equity initiatives and interventions. For example, food expenses could be addressed among all patients, regardless of the presence or absence of traditionally recognized risk factors. This approach is already used (eg, needs-independent school lunch programs) and recognized as having broad positive implications on well-being and equity.^29–32

Financial distress is a multifaceted crisis that impacts patient and family health outcomes. Adults experiencing financial distress reported more anxiety and limited recovery after a critical illness, with more than one-quarter also reporting delaying health care owing to costs.^2,3,10 The importance of understanding similar dynamics in children, particularly modifiable risk factors during hospitalization, is again highlighted by literature demonstrating the adverse effect of PICU admissions on financial status.²⁸ Interventions supporting family needs (eg, nursing follow-up, medical-legal partnerships) are already being trialed,^33,34 but more broadly identifying at-risk families allows for expanded reach of these interventions, potentially magnifying their impact. Although we studied a PICU population, the nature of NOPEs and concerns regarding food expenses are likely generalizable to families of any hospitalized child.¹² Although we may find a similar relationship between NOPEs and financial distress, the magnitude may differ across settings.

In highlighting the subjective cost burden of food, we identified a modifiable factor that could mitigate financial distress, especially in the current climate of increased inflation and prices for necessities. Potential approaches include universal food coupons to hospital cafeterias,³⁵ hospital food banks, and partnerships with community food banks. Beyond instituting changes, we need to leverage perceived financial distress as a relevant outcome to study the efficacy of these interventions and examine the impact of perceived financial distress on post-PICU outcomes (eg, medication adherence, readmission rates, quality of life).

We acknowledge that this study has limitations. This was a single-center study, and distribution of financial distress and NOPEs likely varies among centers. We recognize that electronic survey distribution excludes families without reliable Internet access and that a separate study is needed to evaluate this vulnerable population. NOPEs were reported from memory, making recall bias possible. To mitigate this, we asked for costs over a short period of time but did not independently validate their recall. We asked about transportation costs over the previous 2 days and other NOPEs over the previous 3 days. This was to prevent families from considering the actual cost of transportation to the hospital before admission, but we realize that this could have caused confusion resulting in inaccurate answers and inconsistent timelines. There was a high attrition rate among families who provided consent and completed surveys. Although this could have led to response bias, our distribution of responses show that this did not appreciably change our results. We used approximate income to calculate the proportion of NOPEs. Not using true annual income may have led to an inaccurate representation of the proportion of expenses. Using the highest possible income, however, we conservatively estimated the minimum proportion of weekly income spent on NOPEs. Regarding statistical analyses, we could have analyzed dollars spent conditioned on income. However, doing so implied that a given dollar amount had a fixed effect across income groups, which we believed to be untrue. Families may have been sensitive about their finances, although we attempted to minimize discomfort by surveying them privately. Many expressed a strong desire to openly discuss their financial distress and unexpected expenses. We did not examine the effect of NOPEs on health outcomes but feel that a necessary next step should include ensuring that interventions not only reduce financial distress but also improve health outcomes. Finally, we did not assess for protective factors (eg, social supports) or adaptability in spending behaviors, both of which are important areas for future work.

CONCLUSION

Caregiver-perceived financial distress is associated with the proportion of weekly income spent on total NOPEs and, in particular, the proportion spent on food. Perceived financial distress might also identify larger groups of financially vulnerable families who may not be captured by SDOH screenings. Given the long-term effects of financial distress, reliably identifying vulnerable families and understanding risk factors for financial distress is crucial to optimizing outcomes.

ABBREVIATIONS

aOR

adjusted odds ratio

CCC

complex chronic condition

COI

Child Opportunity Index

COL

Child Opportunity Level

IFDFW

InCharge Financial Distress/Financial Well-Being

NOPE

nonmedical out-of-pocket expense

NUCATS

Northwestern University Clinical and Translational Science Institute

OOP

out-of-pocket

PELOD

Pediatric Logistic Organ Dysfunction

PICU

pediatric intensive care unit

REDCap

Research Electronic Data Capture

SDOH

social determinants of health

SES

socioeconomic status