1 INTRODUCTION

Half a million childhood cancer survivors are living in the United States (2020) with significant improvement in survival and other health outcomes.¹ However, two-thirds of these survivors experience late physical and emotional effects from the treatment for their childhood cancer (e.g., chemotherapy and radiation), which increases their healthcare needs.^2-4 It is important to understand long-term healthcare utilization and access among childhood cancer survivors (CCS) to facilitate timely public health programs,⁵ with the ultimate goal of improving clinical health outcomes and quality of life for this vulnerable population.

A few studies have investigated patterns of healthcare utilization, barriers to healthcare access, and their associated factors among CCS.^{6, 7} Cousineau et al. showed that insurance coverage for CCS was a key driver to seeking care from cancer specialists and primary care physicians.⁶ Using a large data set from the Childhood Cancer Survivor Study (CCSS), Mueller et al. showed that health insurance increased the likelihood of outpatient care.⁷ Baedke et al. using the St. Jude Life data found that a quarter of CCS (24.8%, 985/3,964) reported “forgoing needed medical care in the last year due to finances.”⁸ This study accounted for age, sex, treatment era, treatment type, primary cancer types, and self-reported health perceptions. There are additional factors that may be related to forgone needed medical care but were not addressed in previous studies.

Andersen's Behavioral Model of Health Services Use⁹ has been widely utilized to assess relationships between healthcare utilization and predisposing, enabling, and need factors. This model suggests additional factors beyond those examined in previous studies may be related to forgone needed medical care. These additional factors may include marital status, residency as predisposing factors, employment as an enabling factor, and health risk behaviors (including smoking, binge drinking, and physical activity) and body mass index as need factors.^{9, 10}

In this paper, we look for additional factors that may affect the use of healthcare by CCS. We used national survey data to evaluate healthcare utilization and cost barriers (i.e., having a checkup, receiving influenza vaccine, or experiencing healthcare cost barriers in the past 12 months) among U.S. CCS by themselves and in comparison with noncancer controls. Using a nationally representative data set has the potential to provide more generalizable findings. The inclusion of unstudied predictors of healthcare utilization and barriers would further knowledge on this topic.

2 METHODS

3 RESULTS

3.2 Comparison of healthcare utilization and cost barriers between CCS and controls

One-third (34.6%) of CCS had received influenza vaccine and two-thirds (63.6%) had a checkup. Controls had similar rates of having received the influenza vaccine (35.8%) and a checkup (60.6%). However, CCS experienced a higher proportion having healthcare cost barriers (25.1% vs. 13.9%, p = 0.001; Figure 1).

3.2.1 Univariate conditional logistic analyses on receiving influenza vaccine, having a checkup, and experiencing healthcare cost barriers among all samples

Respondents in lower income groups were more likely to have healthcare cost barriers (odds ratio [OR] range by group, 2.52−4.21) compared with those in the highest income group (p < 0.001). Compared with respondents with healthcare coverage, those without any coverage had lower likelihood of receiving influenza vaccine (OR = 0.14; 95% confidence interval (CI): 0.07−0.28), lower likelihood of having a checkup (OR = 0.27; 95% CI, 0.16−0.43), and higher likelihood of experiencing healthcare cost barriers (OR = 6.04; 95% CI, 3.37−10.80; Table 2).

TABLE 2. Results from univariate conditional logistic regression analyses assessing factors associated with healthcare utilization and cost barriers for all samples (N = 923).

	Influenza vaccine		Checkup with a doctor		Healthcare cost barrier
	Crude OR (95% CI)	p value	Crude OR (95% CI)	p value	Crude OR (95% CI)	p value	Healthcare cost barrier Adjusted OR (95% CI)	p value
Demographics
All sample
Childhood cancer survivors	0.95 (0.69-1.30)	0.741	1.17 (0.86 - 1.59)	0.323	2.08 (1.44 - 3.01)	<0.001***	1.72 (1.11-2.65)	0.015
Controls	Reference		Reference		Reference		Reference	*
Sex, N (%)							–	–
Female	1.62 (1.17-2.25)	0.004**	1.62 (1.20-2.19)	0.002**	1.63 (1.05 - 2.52)	0.031*
Male	Reference		Reference		Reference
Education Level, N (%)				0.772				0.099
aBelow college level	0.30 (0.19-0.45)	< 0.001***	0.95 (0.66 - 1.36)		1.76 (1.14 - 2.69)	0.010*	0.63 (0.36-1.09)
Some college or technical school/College graduate	Reference		Reference		Reference		Reference
Employment, N (%)
bOut of work/a homemaker	0.50 (0.30-0.83)		1.15 (0.71-1.86)		1.15 (0.65-2.05)		0.89 (0.46-1.74)
A student/retired	0.95 (0.51-1.78)	0.060	1.36 (0.75-2.48)	0.043*	0.69 (0.29 - 1.62)	0.044*	0.42 (0.18-1.27)	0.466
Unable to work	0.71 (0.35-1.47)		2.92 (1.35-6.32)		2.98 (1.32-6.72)		0.64 (0.21-1.91)
Employed/self-employed	Reference		Reference		Reference		Reference
Marital, N (%)
Never married	0.89 (0.59-1.35)		0.95 (0.63 - 1.42)		1.51 (0.92-2.49)		1.73 (0.97-3.11)
Divorced/widowed/separated	0.59 (0.35-1.00)	0.146	0.81 (0.49 - 1.35)	0.714	2.23 (1.23 - 4.02)	0.018*	1.08 (0.52-2.25)	0.178
Married/A member of an unmarried couple	Reference		Reference		Reference		Reference
Do you have any healthcare coverage? N (%)
No	0.14 (0.07-0.28)	<0.001***	0.27 (0.16 - 0.43)	<0.001***	6.04 (3.37 - 10.80)	<0.001***	6.11 (3.04-12.28)	<0.001***
Yes	Reference		Reference		Reference		Reference
Income level, N (%)
< $15,000	0.43 (0.23-0.80)		0.82 (0.44-1.52)		4.21 (2.09-8.48)
$15,000-$25,000	0.49 (0.30-0.81)		0.53 (0.33-0.85)		3.58 (1.93-6.63)
$25,000-$35,000	0.42 (0.23-0.75)		0.32 (0.18-0.55)		2.52 (1.24-5.13)
$35,000-$50,000	0.59 (0.35-1.01)		0.72 (0.42-1.24)		3.45 (1.79-6.66)
Don't know/not sure/missing	0.49 (0.27-0.90)	0.002**	0.43 (0.24-0.77)	<0.001***	2.01 (0.89-4.56)	<0.001***
$50,000 or more	Reference		Reference		Reference
Health risk behaviors
Smoking status, N (%)
s Current smokers	0.43 (0.29-0.66)	<0.001***	0.53 (0.36-0.77)	0.001***	2.93 (1.87-4.57)	<0.001***	2.16 (1.25-3.74)	0.006**
Nonsmokers	Reference		Reference		Reference		Reference
Do you binge drink? N (%)							–	–
Yes	0.85 (0.58-1.26)	0.425	0.61 (0.41-0.90)	0.012*	1.33 (0.83-2.13)	0.244
No	Reference		Reference		Reference
Physical activity level, N (%)
Not Active	1.20 (0.80-1.80)	0.377	0.98 (0.66-1.46)	0.918	2.19 (1.35-3.54)	0.001**	2.00 (1.12-3.58)	0.019
Active	Reference		Reference		Reference		Reference
Self-reported health status, N (%)
Fair or poor health	0.59 (0.36-0.96)	0.035*	1.27 (0.81-2.00)	0.302	4.31 (2.56-7.25)	<0.001***	2.90 (1.46-5.75)	0.002**
Good or better health	Reference		Reference		Reference		Reference
Days with poor mental health, N (%)
1-13 days mental health not good	1.04 (0.73-1.47)	0.242	0.95 (0.67-1.34)	0.844	2.96 (1.79-4.92)	<0.001***
14+ days mental health not good	0.69 (0.43-1.10)		0.88 (0.55-1.39)		6.18 (3.46-11.06)
Zero day mental health not good	Reference		Reference		Reference		–	–
Days with poor physical health, N (%)
1-13 days when physical health not good	0.82 (0.56-1.18)	0.414	1.02 (0.71-1.48)	0.986	2.71 (1.83-4.03)	<0.001***
14+ days when physical health not good	0.78 (0.48-1.28)		1.04 (0.64-1.67)		3.85 (2.38-6.24)
Zero day when physical health not good	Reference		reference		Reference
Chronic conditions
Body mass index, N (%)
Overweight/obesity	1.16 (0.82-1.64)	0.464	0.95 (0.686-1.35)		1.34 (0.86-2.09)	0.367
Unknown	0.79 (0.39-1.61)		1.03 (0.51-2.06)	0.952	0.98 (0.43-2.20)
Under/normal weight	Reference		Reference		Reference
Number of chronic conditions
>1	0.75 (0.51-1.11)	0.145	0.96 (0.66-1.41)	0.845	4.41 (2.79-6.99)	<0.001***
None or 1	Reference		Reference		Reference

• Models were adjusted for year of data, age, race/ethnicity, residency region, and number of children in the household.
• * p < 0.05.
• ** p < 0.01.
• *** p < 0.001.
• ^a Below college level includes those who never attended school or only attended kindergarten, those who attended only elementary school or some high school, and those who were high school graduates.
• ^b Out of work includes those who were out of work for one year or more and those who were out of work for less than one year.

An educational attainment below college level, being divorced/widowed/separated, having no healthcare coverage, lower income, being physically inactive, reporting fair or poor health, having poor mental health for ≥1 day, having poor physical health for ≥1 day, and having > 1 chronic health condition were associated with increased likelihood of having healthcare cost barriers (OR range by covariate, 1.76-6.18, p  < 0.05). Being current smokers was associated with a lower likelihood of receiving influenza vaccine or having a checkup and with a higher likelihood of experiencing healthcare cost barriers. CCS were more likely to experience healthcare cost barriers (OR = 2.08; 95% CI, 1.44−3.01) compared with controls (Table 2).

3.2.2 Multivariable conditional logistic analysis on experiencing healthcare cost barriers among all samples

CCS remained more likely to experience healthcare cost barriers (adjusted OR [aOR] = 1.72; 95% CI, 1.11-2.65) compared with controls. Respondents without healthcare coverage had a six-fold likelihood to experience healthcare cost barriers (aOR = 6.11; 95% CI, 3.04-12.28). CCS being current smokers were twice likely to experience healthcare cost barriers (aOR = 2.16; 95% CI, 1.25-3.74). Because CCS had a much higher proportion of having multiple chronic health conditions than controls, the covariate of being a CCS and the count of chronic health conditions were highly correlated. Thus, our primary multivariable model excluded the latter. In a separate analysis including both covariates (Supporting Information), the count of chronic condition was statistically significant while the status of being a CCS was not.

4 DISCUSSION

Compared with controls, CCS had lower income, lower educational attainment, more chronic health conditions, and less healthcare coverage. Also, CCS had higher proportions of being current smokers, reporting worse mental health and worse physical health. CCS had similar proportions of having a checkup and receiving influenza vaccine but a higher proportion of experiencing healthcare cost barriers than controls. Among all samples, those with multiple chronic conditions were more likely to face cost barriers. With a higher proportion of having multiple chronic conditions (98.4% vs 94.4%), CCS had a higher chance to face cost barriers in comparison with controls. Among CCS alone, being 25-29 years, no healthcare coverage, and having multiple chronic health conditions were associated with higher likelihood of experiencing healthcare cost barriers. CCS who were current smokers were less likely to have a checkup compared with nonsmoker counterparts.

Our childhood cancer survivor samples shared similarities with and differed from samples in previous studies. Most of our CCS were White, similar to the current literature.^{20, 21} However, most of our survivors were female (70%) compared with 54.5% in Kirchhoff et al. study.²² Our study had more uninsured survivors (18.2%) than previous studies (9.3%-10.2%).^{23, 24} Our findings on CCS having worse socioeconomic outcomes and worse mental and physical health confirmed previous findings.²⁵ The proportion of current smokers among our survivors was 30.3%—higher than previous findings (i.e., 22% prevalence of CCS being smokers in a meta-analysis²⁶ and a current smoking rate of 19% in a study using the CCSS data).²⁷ Of note, the survivors in our sample reported a later age of childhood cancer diagnosis compared with previous studies using the CCSS data.^{3, 28, 29}

U.S. cancer patients face serious healthcare cost barriers.³⁰ Compared with their noncancer siblings, CCS pay more out-of-pocket medical costs.²⁴ We found that CCS had a higher likelihood to experience healthcare cost barriers and had a higher prevalence of being uninsured compared with the general population counterparts. Our finding that a quarter of CCS reported healthcare cost barriers corroborates Beadke et al. (24.8% forgoing needed medical care due to finances).⁸ In our study, the proportions of having a checkup (63.6%) and an influenza vaccine (34.6%) among CCS agreed with previous studies.^{6, 15} Our finding that the proportions having a checkup and an influenza vaccination were similar for survivors and controls indicates equal access to primary care between the two groups. However, CCS had almost doubled the proportion experiencing healthcare cost barriers compared with controls even after adjustments with a multivariable model. This indicates unmet need by CCS for intensive long-term follow-up services beyond primary care (i.e., appointments with cancer specialists and preventive care services starting at earlier ages).²

We found that an absence of healthcare coverage was an important factor associated with healthcare cost barriers among CCS. U.S. adult survivors of childhood cancer encounter fragmented and suboptimal healthcare due to multiple issues related to health insurance hardship (e.g., change in employer-sponsored plans, substantial deductibles, and high healthcare costs).³¹ Addressing healthcare coverage is key to improving long-term follow-up care and, ultimately, health outcomes among CCS. Special programs that guarantee basic needs of healthcare to CCS will be helpful. Our findings showed no impact of the Medicaid expansion under the Affordable Care Act on the healthcare utilization of CCS, which could be explained by a relatively small sample size. Medicaid expansion could be among the most potential programs that provide fundamental healthcare coverage (i.e., checkups with specialists) for CCS.^{6, 32} We suggest states with existing and future Medicaid expansion consider higher emphasis on including specific programs supporting CCS.

We found an increased likelihood of experiencing healthcare cost barriers among CCS aged 24-29 years in comparison with the youngest group (18-24 years). Meanwhile CCS of other age groups had a similar experience of healthcare cost barriers when compared with the youngest group. This finding may indicate an interrupted healthcare access when adult survivors of childhood cancer transition out from parents’ healthcare plans.^{33, 34} This finding also corroborates the fact that U.S. population aged 18-29 years have the highest proportion of noninsurance compared with all other age groups (30-64 years).³⁵ Thus, we advocate for programs effectively addressing interrupted care across transitions for CCS aged 24-29 years^36-38 based on the insurance disadvantage. This recommendation has contributed to the current literature that has suggested pretransition and transition planning based on CCS’ clinical needs and efficacy/comfort of specialty and primary care physicians in managing chronic health conditions connected to childhood cancer.^39-41

The status of being current smokers was more prevalent among CCS than controls. Also, among CCS, being current smokers was associated with lower utilization of checkups with doctors. Meanwhile, CCS are at elevated risk for developing subsequent illnesses including cardiovascular diseases, second cancer, and pulmonary complications.⁴² A major cause for all these three illnesses,⁴³ smoking itself is harmful to the health of survivors and can exacerbate late effects from the treatment for primary cancer.^{44, 45} Our findings show that the status of being a current smoker creates another layer of harm to survivors by decreasing healthcare utilization in addition to, previously well-known, smoking's health-worsening harm. We suggest the importance of developing tailored smoking cessation interventions for CCS.

Our study has several limitations. Although BRFSS survey data aim to represent the nation, a large portion of our childhood cancer survivor sample was from the Midwest. We lack important cancer-related information (e.g., stage of the primary cancer, type and site of cancer, type of received treatments). Additionally, our study is subject to several other limitations inherent to the BRFSS survey design, as reported by a previous study.²⁵ These limitations include the inability to study households without phone lines, the response rate possibly affecting the representativeness (also known as nonresponse bias), and a small sample size of the survivors.

Our study's strengths include CCS and their matched noncancer counterparts, which enabled us to evaluate the impact of being a childhood cancer survivor and other factors on healthcare utilization and cost barriers. The BRFSS survey's recruitment method is not focused on hospital-based populations, offering a less biased sample (hospitals tend to recruit patients with severe health issues). Our findings help further knowledge on childhood cancer survivorship research by inclusion of unstudied predictors—marital status, employment, health-related risk behaviors, which could be associated with healthcare utilization and cost barrier in CCS—a population with continuously improved life expectancy. We used sample aged < 50 years because chronic health conditions are more prevalent in older age groups and increased use of healthcare services is also more commonly seen in aging populations as seen in a previous study.²⁵ This age cutoff helped strengthen the opportunity to contrast healthcare utilization and cost barriers by the status of being CCS. Furthermore, our analysis included the assessment of health risk behaviors, clinical factors as need factors, and several predisposing factors that have never been studied in healthcare utilization among U.S. CCS. These focused sample and comprehensive sets of variables enhanced the ability of our study to make recommendations for long-term follow-up care for CCS.

In conclusion, CCS are more likely to encounter healthcare cost barriers compared with their general population counterparts. Among CCS, being 25−29 years, uninsured, or a current smoker is associated with poorer outcomes in healthcare utilization and access. We advocate for effective public health programs addressing care transitions and smoking cessation for this vulnerable group.

AUTHOR CONTRIBUTIONS

Study design (Nghiem, Alanaeme, Wong); data acquisition, analysis, and interpretation (Nghiem, Alanaeme, Wong, Mennemeyer); drafting (Nghiem, Wong, Alanaeme, Mennemeyer); final approval of the version to be published (Nghiem, Wong, Alanaeme, Mennemeyer)

CONFLICTS OF INTEREST STATEMENT

The authors report no other conflicts of interest.