Rural-Urban Differences in Health Care Unaffordability during the Postpartum Period

Abstract

Objective.

To examine health care unaffordability for rural and urban residents and by postpartum status.

Methods.

We used cross-sectional survey data on female identifying respondents ages 18–44 (n=17,800) from the 2019–2021 National Health Interview Study (NHIS). Outcomes of interest were three measures of health care unaffordability. We conducted bivariate and multivariable regression models to assess the association between health care unaffordability, rurality, and postpartum status.

Results.

Bivariate analyses showed postpartum people reported statistically significantly higher rates of being unable to pay medical bills and having problems medical paying bills, as compared to non-postpartum people. Rural residents also reported statistically significantly higher rates of being unable to pay their medical bills and having problems paying medical bills as compared to urban residents. In adjusted models, the predicted probability of being unable to pay medical bills among postpartum respondents was 12.8 percent (CI: 10.1–15.5), which was statistically significantly higher than among non-postpartum respondents. Similarly, postpartum respondents had statistically significantly higher predicted probabilities of reporting problems paying medical bills (18.4 percent, CI: 15.4–21.4) as compared to non-postpartum respondents. Rural residency was not significantly associated with the health care unaffordability outcome measures in adjusted models.

Conclusions.

Both postpartum and rural respondents reported higher rates of being unable to pay medical bills and having problems paying medical bills, however after adjusting for covariates, only postpartum respondents reported statistically significantly higher rates of these outcomes. These results suggest that postpartum status may present challenges to health care affordability that span the urban/rural context.

Introduction

In the United States, rising health care costs and high need for health care services have left millions with medical debt.¹ As of 2019, overall medical debt was more likely to be experienced in the southern United States, in states that did not expand Medicaid under the Affordable Care Act, and in rural areas.² Previous research has found high out-of-pocket costs associated with health care in the period immediately before and after birth with out-of-pocket costs for maternity care increasing over time.³ This increase has been attributed to rising deductibles,³ rising maternal morbidities which require more expensive health care encounters,⁴ and lack of Medicaid access, which has proven protective against high out-of-pocket costs in the postpartum period.⁵ Moreover, maternal complications like severe maternal morbidity, which is more common among rural residents, have been linked to increased health care costs by almost 75% for childbirth and related readmissions.^6,7 As a result, recent estimates have reported that nearly one-quarter of postpartum people describe worries regarding paying their medical bills from childbirth, with 8.3% of postpartum people reporting bills from childbirth being sent to collection agencies.⁸ This study adds to existing research by estimating health care unaffordability at the intersection of rural residency and postpartum status. We hypothesize that the compounding effect of living in a rural area and being postpartum is associated with reporting higher health care unaffordability.

Methods

Data and Sample

In this cross-sectional analysis, we used 2019–2021 National Health Interview Survey (NHIS) data, accessed through the Integrated Public Use Microdata (IPUMS) Health Surveys at the University of Minnesota.⁹ The NHIS is a nationally representative survey of the civilian, noninstitutionalized population in the U.S. We selected 2019–2021 survey years to capture the experiences of rural and postpartum people using the most recent data available, pooled across three separate survey years for adequate sample sizes for examining postpartum individuals and urban-rural residency. Furthermore, from 2019–2021 the NHIS used the same survey instrument and sampling design, which was not true prior to the 2019 NHIS redesign.⁹ In addition, 2019 was the first year NHIS made the rural/urban variable used in this analysis available in the public use database. Our sample included female identifying respondents of reproductive age 18–44 (n=17,800). Male identifying respondents were excluded from the sample as only female identifying respondents were asked whether they had a live birth in the past 12 months.

Variables

Our outcomes of interest were reports of health care unaffordability. To understand the complexity of health care unaffordability, this study examined responses to three NHIS questions capturing worries about medical bills, problems paying medical bills, and being unable to pay medical bills. Worry about medical bills was dichotomized into categories of very worried or somewhat worried versus not at all worried. Problems paying medical bills was already dichotomized as a yes or no question within the original survey. Respondents were only asked about being unable to pay medical bills if they responded “yes” to the question of having problems paying bills. Being unable to pay bills was also already dichotomized into categories of yes or no in the original survey design.

The key independent variables of interest for this study were postpartum status and rurality. Respondents were considered postpartum if they responded yes to having a live birth in the past 12 months. Respondents were classified as rural if they resided in rural/non-metropolitan counties or urban if they resided in a metropolitan county. Non-metropolitan and metropolitan classifications were based on the 2013 NCHS Urban-Rural Classification Scheme and provided in the NHIS data.¹⁰

We included several sociodemographic variables in our analyses including age, income (measured through the income to poverty ratio variable), race and ethnicity (non-Hispanic white, Hispanic, Black, Asian, American Indian/Alaskan Native), maximum educational attainment of all adults in the household (less than high school, high school/general education development (GED), some college/associates degree, four-year college graduate), marital status, and health insurance type (no health insurance, private insurance, Medicaid). Lastly, we included a variable describing number of children in the household to understand whether health care unaffordability is related to having children in general or if the postpartum period had its own unique relationship with health care unaffordability. These sociodemographic covariates were selected based on research indicating that medical debt is disproportionately shouldered by certain communities including Black Americans and those who are low-income.² Indicators for missingness were created for all covariates with missing data. All outcome variables, key independent variables of interest and sociodemographic covariates were included and consistent across all three years of survey data.

Analyses

We calculated weighted sociodemographic characteristics at the intersection of postpartum status and rural residence. Strata and PSU variables were used for all analyses in accordance with NHIS recommendations and sample years 2019–2020 were combined following NCHS guidance.¹¹

We calculated the prevalence of each outcome variable (worry about paying medical bills, problems paying medical bills, and unable to pay medical bills) overall, by rural versus urban residency, by postpartum status, and by rural versus urban residency among postpartum respondents using unadjusted survey-weighted proportions. Survey weighted chi-squared tests were used to test for bivariate differences.

Finally, we used multivariable logistic regression models to examine the role of sociodemographic variables in explaining the relationship between rural-urban residency and postpartum status with the health care unaffordability outcome variables. These models included covariates for sociodemographic variables and year fixed effects, as well as a residency-postpartum status interaction. Using Stata’s margins command, we then conducted post-estimation analyses of our logistic regression models to obtain adjusted predicted probabilities. Adjusted odds ratio results are available in Appendix Table 1. Pooled survey weights were created based on NHIS recommendations for combining multiple years of data and were used in all statistical analyses.⁹

Results

Table 1 describes the sociodemographic characteristics of the population by postpartum and rural status. The study included 17,800 total respondents, weighted to represent 57,927,729 individuals. The weighted percentage of the postpartum population was 6.24% and the weighted percentage of the rural population was 11.60%. We then compared sample characteristics for four subpopulations of the sample: rural postpartum respondents (0.09%), rural not postpartum respondents (11.16%), urban postpartum respondents (5.20%) and urban not postpartum respondents (81.55%). Rural postpartum respondents were younger (60.56% age 25–34), majority non-Hispanic white (77.53%), more likely to be married (66.70%), most likely to have a maximum household education of High School/GED (30.67%), and were most likely of all the groups to have three or more children (37.04%). In addition, rural postpartum respondents were least likely to have private health insurance of the four groups and most likely to have Medicaid insurance (31.3%).

Table 1.

Sample Characteristics, National Health Interview Survey 2019–2021

Characteristics	Rural Postpartum (N=163) 0.09%	Rural Not Postpartum (N=1,987) 11.16%	Urban Postpartum (N=925) 5.20%	Urban Not Postpartum (N=14,516) 81.55%
Age
18–24	26.58	23.45	15.90	25.97
25–34	60.56	37.47	59.94	37.16
35–44	12.87	39.07	24.16	36.88
Ratio Income to Poverty Threshold
<1.0	22.98	20.89	20.78	13.29
1.0–1.99	30.88	26.45	22.10	19.44
2.0–2.99	20.10	18.90	14.63	17.38
3.0–3.99	11.87	12.51	11.92	13.51
4.0–4.99	7.24	9.00	7.92	10.71
5.0+	6.92	12.25	22.64	25.66
Race/Ethnicity
Non-Hispanic white	77.53	73.33	49.14	51.74
Hispanic	14.86	9.57	26.70	23.45
Black	6.26	8.97	14.41	14.64
Asian	0.17	1.55	6.79	7.33
AI/AN	1.01	5.48	0.85	1.07
Marital Status
Unmarried	33.30	55.75	37.53	61.12
Married	66.70	44.25	62.47	38.87
Number of Children
0	4.33	34.02	2.31	46.66
1	26.86	24.41	34.86	20.76
2	31.56	22.10	33.95	20.07
3+	37.04	19.34	28.75	12.37
Health Insurance Type
No Health Insurance	17.47	17.32	11.38	13.55
Private Health Insurance	45.95	53.91	54.60	65.00
Medicaid Insurance	31.3	23.72	25.15	14.87
Other Coverage	5.27	4.91	8.87	6.19
Education
< High School	6.82	7.07	6.21	38.00
High School/GED	30.67	22.70	20.87	16.96
Some College/Associates	34.05	40.07	25.19	31.12
Four Year College Graduate	28.46	30.02	47.73	47.94

Notes: Sample includes female respondents ages 18–44 years. All percentages are weighted using NHIS survey weights.

Abbreviations: AI/AN, American Indian or Alaskan Native; GED, General Education Development

In weighted bivariate analyses, postpartum people were significantly more likely to report two of the three unaffordability outcomes at significantly higher rates than non-postpartum people (Table 2). Postpartum people reported significantly higher rates of being unable to pay their medical bills (11.87% postpartum vs. 9.18% non-postpartum, p<.001) and having problems paying their medical bills (16.92% postpartum vs. 14.06% non-postpartum, p<.001). However, postpartum people reported significantly lower rates of being worried about their medical bills (52.16% postpartum vs. 52.80% non-postpartum p<.001). Rural residents were significantly more likely to report being unable to pay their medical bills (12.43% rural vs. 8.93% urban, p=<.01). Rural residents were also significantly more likely to report having problems paying bills as compared to urban residents (18.33% rural vs. 13.67% urban, p<.01).

Table 2.

Prevalence of Health Care Unaffordability Overall and by Postpartum and Urban-Rural Status, National Health Interview Survey 2019–2021

Variables	Overall	Postpartum Status			Urban-Rural Residence			Postpartum and Urban-Rural
Health Care Unaffordability Measures	Overall Prevalence (n=17,800)	Non- Postpartum (n=16,503)	Postpartum (n=1,088)	p-value	Urban (n=15,633)	Rural (n=2,167)	p-value	Postpartum + Urban (n=925)	Postpartum + Rural (n=163)	p-value
Unable to Pay Medical Bills	9.34	9.18	11.87	<.001	8.93	12.43	<0.01	10.99	13.64	0.74
Problems Paying Medical Bills	14.21	14.06	16.92	<.001	13.67	18.33	<0.01	16.09	16.37	0.93
Worried About Paying Medical Bills	52.62	52.80	52.16	<.001	52.70	52.01	0.93	50.64	48.76	0.85

Notes: Sample includes female respondents ages 18–44 years. Weighted proportions presented using NHIS survey weights. Survey weighted chi-squared tests used to test for bivariate differences.

In adjusted models, postpartum status was associated with a statistically significantly higher predicted probability of being unable to pay medical bills relative to non-postpartum respondents even after adjusting for relevant covariates (12.8%, CI: 10.1–15.5) (Table 3). Similarly, postpartum status was also associated with a statistically significantly higher adjusted predicted probability of having problems paying bills (18.4%, CI: 15.4–21.4) relative to non-postpartum respondents. We also found differences in the adjusted predicted probabilities of health care unaffordability outcomes by race and ethnicity. Hispanic ethnicity was associated with statistically significantly lower adjusted predicted probabilities of being unable to pay medical bills (8.0%, CI: 6.9–9.1) and having problems paying medical bills (12.6%, CI: 11.2–14.0) as compared to the Non-Hispanic white reference group. However, Hispanic ethnicity was associated with a statistically significantly higher adjusted predicted probability of being worried about medical bills (61.2%, CI: 58.9–63.5) relative to non-Hispanic white respondents. In addition, the adjusted predicted probability of being unable to pay medical bills were statistically significantly lower for Asian individuals (3.3%, CI: 1.8–4.7) relative to non-Hispanic white individuals, as was the adjusted predicted probability of having problems paying medical bills (6.4%, CI: 4.5–8.4). Unsurprisingly, compared to those without insurance, having private health insurance, Medicaid, or other coverage was associated with statistically significantly lower adjusted predicted probabilities of being unable to pay medical bills, having problems paying medical bills, and being worried about medical bills. Lastly, compared to 2019, the years 2020 and 2021 were associated with statistically significantly lower adjusted predicted probabilities of being unable to pay medical bills and having problems paying medical bills, pointing toward the unique time frame under investigation.

Table 3.

Multivariable Regression Models for Predictors of Health Care Unaffordability, National Health Interview Survey 2019–2021

Variables	Unable to Pay Medical Bills (N=17,588)	Problems Paying Medical Bills (N=17,588)	Worried About Paying Medical Bills (N=17,588)
	Adjusted Predicted Percentage, (CI)	Adjusted Predicted Percentage, (CI)	Adjusted Predicted Percentage, (CI)
Postpartum Status
Postpartum	12.8*	18.4**	54.8
	(10.1–15.5)	(15.4–21.4)	(50.9–58.7)
Not Postpartum	9.4	14.2	52.8
	(8.8–10.0)	(13.4–14.9)	(51.7–53.9)
Rural Status
Rural	9.8	15.5	51.4
	(8.0–11.7)	(13.3–17.7)	(48.5–54.3)
Urban	9.6	14.3	53.1
	(8.9–10.3)	(13.4–15.1)	(52.0–54.2)
Postpartum*Rural
Postpartum + Rural	13.0	19.7	53.3
	(9.4–16.6)	(15.6–23.8)	(48.3–58.2)
Postpartum + Urban	12.7	18.2	55.0
	(10.0–15.4)	(15.2–21.2)	(51.1–58.9)
Not Postpartum + Rural	9.6	15.3	51.3
	(7.8–11.4)	(13.1–17.4)	(48.4–54.1)
Not Postpartum + Urban	9.4	14.0	53.0
	(8.7–10.1)	(13.2–14.8)	(51.8–54.2)
Age
18–24	6.9	11.4	43.1
	(5.8–8.0)	(10.0–12.9)	(40.9–45.3)
25–34	10.9***	15.5***	55.5***
	(9.9–11.9)	(14.4–16.6)	(54.1–57.0)
35–44	10.5***	15.7***	57.0***
	(9.5–11.5)	(14.5–16.8)	(55.6–58.5)
Ratio Income to Poverty Threshold
<1.0	13.1	18.9	58.4
	(11.2–15.1)	(16.6–21.3)	(55.6–61.2)
1.0–1.99	13.1	19.1	60.9
	(11.6–14.5)	(17.3–20.8)	(58.5–63.2)
2.0–2.99	12.2	18.7	59.7
	(10.6–13.8)	(16.8–20.5)	(57.4–62.0)
3.0–3.99	7.9***	13.2***	55.1
	(6.5–9.3)	(11.5–14.9)	(52.6–57.6)
4.0–4.99	5.9***	10.8***	49.7***
	(4.3–7.6)	(8.9–12.6)	(47.1–52.3)
5.0+	2.6***	5.5***	39.2***
	(1.9–3.3)	(4.5–6.5)	(37.3–41.2)
Race/Ethnicity
Non-Hispanic white	10.3	15.5	50.9
	(9.4–11.2)	(14.4–16.6)	(49.6–52.2)
Hispanic	8.0**	12.6**	61.2***
	(6.9–9.1)	(11.2–14.0)	(58.9–63.5)
Black	11.7	16.4	49.4
	(10.2–13.3)	(14.6–18.3)	(46.7–52.2)
Asian	3.3***	6.4***	53.0
	(1.8–4.7)	(4.5–8.4)	(49.2–56.8)
AI/AN	12.8	15.9	44.1
	(6.1–19.4)	(8.9–22.9)	(34.3–52.3)
Education
< High School	8.9	13.0	54.4
	(6.6–11.3)	(10.0–15.9)	(49.7–59.0)
High School/GED	11.4	16.0	52.9
	(10.1–12.8)	(14.4–17.6)	(50.0–55.4)
Some College/Associates	10.8	16.0	54.5
	(9.8–11.8)	(14.8–17.1)	(52.9–56.2)
Four Year College Graduate	7.1	12.2	51.7
	(6.1–8.1)	(11.1–13.4)	(50.2–53.3)
Marital Status
Unmarried	10.3	15.1	53.7
	(9.4–11.1)	(14.1–16.1)	(52.4–55.0)
Married	8.5**	13.3*	51.8
	(7.5–9.4)	(12.1–14.5)	(50.2–53.5)
Health Insurance Type
No Health Insurance	16.1	21.0	76.2
	(14.1–18.1)	(18.7–23.2)	(73.8–78.7)
Private	8.3***	14.0***	53.2***
	(7.5–9.1)	(13.0–15.0)	(51.8–54.5)
Medicaid	8.4***	11.6***	41.2***
	(7.2–9.6)	(10.2–13.1)	(38.6–43.7)
Other Coverage	7.0***	11.2***	36.6***
	(5.4–8.7)	(9.0–13.4)	(33.3–39.9)
Number of Children
0	10.5	14.8	55.4
	(9.4–11.6)	(13.6–16.1)	(53.9–57.0)
1	9.9	14.7	53.2
	(8.7–11.0)	(13.3–16.2)	(51.3–55.3)
2	9.2	14.7	50.5***
	(8.0–10.4)	(13.3–16.2)	(48.5–52.6)
3+	8.1**	12.8	48.2***
	(6.8–9.4)	(11.0–14.5)	(45.8–50.6)
Year
2019	10.8	16.2	52.5
	(9.8–11.7)	(15.1–17.4)	(50.9–54.0)
2020	9.0*	14.1**	54.8*
	(8.0–10.1)	(12.9–15.3)	(53.1–56.4)
2021	9.0*	12.9***	51.6
	(8.0–10.1)	(11.8–14.1)	(50.0–53.2)

Notes: Sample includes female respondents ages 18–44 years. Weighted adjusted predicted probabilities presented using NHIS survey weights.

^*p<0.05,

^**p<0.01,

^***p<0.001

Abbreviations: AI/AN, American Indian or Alaskan Native. GED, General Education Development. CI, 95% confidence interval.

Discussion

This study examined the relationships between rurality, postpartum status, and healthcare unaffordability. We initially posited that the intersectional identity of being rural and postpartum would lead to greater healthcare unaffordability; however, our results paint a more complicated picture. While findings show postpartum people separately report significantly higher rates of being unable to pay for bills and having problems paying for bills, rural status was only significantly associated with these factors in unadjusted analyses; results were no longer statistically significant after adjustment for relevant sociodemographic variables. Further, we did not find significant compounding effects at the intersection of postpartum status and rurality when examining the interaction of those factors. These findings could point toward healthcare unaffordability in the postpartum period being an issue regardless of rural/urban location and further supports existing advocacy and legislation to expand health coverage in the postpartum period especially for those who are low-income.^5,12

When interpreting these findings, it is important to bear in mind temporal and geographic contexts. The topic of healthcare unaffordability was particularly salient in the time frame under investigation due to the COVID-19 pandemic and its accompanying legislation. Indeed, health care unaffordability concerns in these years, particularly for postpartum people, were affected by both pandemic relief measures and increased postpartum insurance coverage, which may have been protective against medical debt accumulation.^13,14

Our findings showed differences in the experience of health care unaffordability by race and ethnicity. For example, Hispanic ethnicity was associated with statistically significantly higher adjusted predicted probabilities of worry about medical bills but significantly lower predicted probabilities of being unable to pay bills and having problems paying bills. These associations demonstrate the potential anxiety produced by accessing health care in the United States where actual cost of services billed to patients remains hidden. This anxiety may be more acutely felt by certain populations (such as respondents of Hispanic ethnicity) due to a historically high risk of being uninsured as compared to other racial/ethnic groups,¹⁵ and may reflect fear about participation in programs such as Medicaid and Children’s Health Insurance Program (CHIP) following Trump-era public charge rules, which have since been reversed by the Biden administration.¹⁶ When examining race and ethnicity, we also found respondents who identified as Asian reported statistically significantly lower adjusted predicted probabilities of worry about medical bills and having problems paying medical bills, which is consistent with recent literature on low rates of medical indebtedness faced by Asian communities.¹⁷

In addition to differences by race and ethnicity, one surprising finding included lower adjusted predicted probabilities of problems paying medical bills experienced by individuals who had one, two, or three or more children compared to those who had no children. One might surmise that based on the high costs of raising children (whether that is higher grocery bills, more medical visits, cost of childcare) that parents of children may experience more difficulty paying medical bills as compared to their peers with no children. However, having children may lead to eligibility in safety net programs, such as CHIP, which could lessen the burden of medical debt.

Lastly, geographic context was not only a predictor variable of interest, but it may also be an important lens for understanding responses. Past studies have demonstrated health care unaffordability is not adequately captured solely by the dollar amount of medical debt reported by collections agencies.¹⁸ Rather, studies have shown that comprehensive questions pertaining to credit card debt and rationing other needed services provide a fuller picture health care affordability¹⁸ Therefore, by examining level of “worry” and identification of “problems” with medical debt, this study expanded on existing literature to capture the experience of health care unaffordability through survey response. While understanding experience is useful in gaining a more complete picture of the issue, it is also limited by the subjective interpretation of respondents. For example, rural and urban residents may differ in their attitudes toward debt and seeking out health care.^19–21

Limitations

Limitations in this study include a lack of focus on varying health factors that could shape the postpartum experience and subsequent medical bills. To account for this concern, a sensitivity analysis was run looking at results stratified by self-rated health and found no significant differences in outcomes. This study is also limited to three years due to availability of variables of interest. However, these three years present a unique time frame in the COVID-19 pandemic during which health care unaffordability could be of particular importance. Additionally, the health care unaffordability questions in the NHIS relate to health care costs generally and are not specific to perinatal or postpartum care, thus we were unable to capture healthcare unaffordability specific to these types of care. However, as evidenced by the push to extend Medicaid coverage to one year postpartum,⁵ it is now recognized that postpartum care is ongoing and may require office visits and follow up care that can last a year or longer, especially to address chronic conditions, family planning needs, and/or mental health concerns.²² Finally, responses to the NHIS self-reported health care unaffordability variables likely presume that individuals have accessed health care and have received medical bills. Given this, our results may be an underestimate as those with no bills because of a lack of access may not report concerns. However, recent literature has called for new ways to capture the experience of health care unaffordability,¹⁸ as previous reports of medical debt in collections fails to capture the broad experience of health care unaffordability and self-report is one, albeit imperfect, way to understand the experience of unaffordability.

Conclusions

This study contributes to the growing body of literature examining health care unaffordability among rural residents and postpartum people. Our findings bolster existing evidence on the critical importance of extending Medicaid coverage for postpartum people while also noting that even those with health insurance face health care unaffordability issues, especially in the postpartum period. Policies to improve health care in rural areas and during the postpartum period should consider the intersection of these backgrounds to account for nuanced challenges and strengths.