Burden and Consequences of Financial Hardship From Medical Bills Among Nonelderly Adults With Diabetes Mellitus in the United States

Introduction

Diabetes mellitus is a growing public health and economic challenge in the United States, affecting 30.3 million individuals¹ and with a continuously increasing healthcare expenditure for its management.^2–4 The total direct costs of diabetes mellitus in the United States have doubled over the last decade from $116 billion in 2007 to $237 billion in 2017,⁵ accounting for a quarter of total healthcare spending, and this trend is driven mainly by the medications and per-encounter costs.⁶ Notably, with an average annual out-of-pocket (OOP) spending of over $1800 and a total lifetime expenditure excess of >$100 000 compared with those without diabetes mellitus,⁷ nearly 1 in 4 individuals with diabetes mellitus are part of a family that spends >10% of their income on healthcare-related expenses.⁸ This high financial burden from OOP costs represents a serious risk of medical hardship for Americans with diabetes mellitus—especially the most vulnerable, such as low socioeconomic status and uninsured—is a barrier to initiate and adhere to their treatment, and limits the person’s capability to cope with unexpected expenses for medical events.^9,10

However, there are no estimates of the recent prevalence and possible consequences of financial hardship from medical bills for individuals with diabetes mellitus in the United States.¹¹ Although diabetes mellitus is a great example of a chronic high-cost condition, it has some particularities that might put patients with this condition on a potential higher risk of financial hardship, such as the need of uninterrupted access to medications and equipment (eg, glucometer, syringes, insulin pump, among others). Knowing the possible unintended medical and nonmedical tradeoffs that patients with diabetes mellitus experience could inform the design and implementation of policies (including the regulation of high-deductibles and cost-sharing strategies) aimed at preventing and mitigating the financial toxicity from medical care.

As a result, we aimed to describe in detail the national scope of financial hardship from medical bills among nonelderly individuals with diabetes mellitus using nationally representative data, and its impact on financial distress, food security, cost-related medication nonadherence (CRN), and foregone/delayed medical care. We also compared these measures with nonelderly individuals without diabetes mellitus and elderly individuals with diabetes mellitus. Although this project’s focus is on patients with diabetes mellitus, the results from our study might be extended beyond this condition to the experiences of patients with other chronic high-cost diseases.

Methods

Data Source

We used the National Health Interview Survey (NHIS) data from 2013 to 2017. The NHIS is a series of annual cross-sectional national surveys that provide information on the health of the noninstitutionalized population of the United States. The sample design follows a multistage area probability design, which adjusts for nonresponse and further allows for national representative sampling of households and individuals, including underrepresented groups.¹² This survey consists of a questionnaire divided into 4 cores: household composition, family core, sample child core, and sample adult core. The household composition file collects basic and relationship information about all persons in a household. The family core file collects sociodemographic characteristics, basic indicators of health status, activity limitations, injuries, health insurance coverage, and access to and utilization of health care services. From each family, one sample child and one sample adult are randomly selected to gather more in-depth information for the sample child core and sample adult core, respectively.¹³ In our study, we used the sample adult core files with variables supplemented from the other cores, complemented with demographic and socioeconomic characteristics. Since NHIS data are deidentified and publicly available, this study was exempt from the purview of the Yale University’s Institutional Review Board Committee.¹⁴ The analyses code that supports the findings of this study are available from the corresponding author upon reasonable request.

Study Population

We included individuals from 18 to 64 years old with a self-reported diagnosis of diabetes mellitus by answering “Yes” to the following question: “Other than during pregnancy, have you ever been told by a doctor or other health professional that you have diabetes or sugar diabetes?” if female, or “Have you ever been told by a doctor or other health professional that you have diabetes or sugar diabetes?” if male. This strategy has been used in the past.^15,16 We excluded older adults to focus on the population without universal financial protection from public insurance.

Study Outcomes

Financial Hardship From Medical Bills

Individuals were classified as having financial hardship from medical bills if they responded “Yes” to any of the following 2 questions: “In the past 12 months did you/anyone in your family have problems paying or were unable to pay any medical bills? Include bills for doctors, dentists, hospitals, therapists, medication, equipment, nursing home or home care,” or “Do you/anyone in your family currently have any medical bills that are being paid off over time? This could include medical bills being paid off with a credit card, through personal loans, or bill paying arrangements with hospitals or other providers. The bills can be from earlier years as well as this year.” Furthermore, if answered “Yes” to having problems paying bills, then a follow-up question was asked: “Do you/Does anyone in your family currently have any medical bills that you are unable to pay at all?”^17,18 Considering those who were unable to pay at all a more severe degree of financial hardship, we then divided our study population into 3 mutually exclusive categories to study the stepwise changes in status of financial hardship from medical bills: No Financial Hardship From Medical Bills, Financial Hardship From Medical Bills but Able to Pay, and Currently Unable to Pay Bills at All.

High Financial Distress, Food Insecurity, Cost-Related Medication Nonadherence, and Foregoing Medical Care

We evaluated financial distress using 6 questions regarding the concern about several financial matters, including having enough for retirement, ability to pay medical costs of serious illness or accident, maintaining standard of living, ability to pay costs of usual healthcare, inability to pay normal monthly bills, and inability to pay rent/mortgage/housing costs (Table I in the Data Supplement). All questions had 4 possible answers: (1) “Not Worried at All,” (2) “Not Too Worried,” (3) “Moderately Worried,” and (4) “Very Worried.” A total aggregate score was created, ranging from 6 to 24 with a higher score reflecting greater distress.¹⁹ Participants within the highest quartile were considered as having high financial distress.

Food security in the past 30 days was created based on the 10-item questionnaire as recommended by the US Department of Agriculture Economic Research Service (Table I in the Data Supplement)^13,20 and constructed following the NHIS instructions.²¹ Answers of ≥3 days were considered affirmative in questions about frequency of occurrence in the past 30 days. A raw score ranging from 0 to 10 was calculated, and participants were categorized as follows: 0 to 2 points: food secure; 3 to 5 points: low food security; and 6 to 10 points: very low food security. We then defined food insecurity as having either low or very low food security, in concordance with previous studies.²²

CRN was assessed by the self-reporting of any of the following: skipping medication doses, taking less medicine than prescribed, and delaying filling a prescription with the objective of saving money (Table I in the Data Supplement).

Foregone or delayed medical care was defined based on an affirmative answer to any of the following questions: (in the past 12 months) has medical care been delayed due to cost; needed and did not get medical care due to cost; could not afford prescription medicine, mental health care or counseling, dental care, eyeglasses, to see a specialist or follow-up care (Table I in the Data Supplement).^23,24

Covariates

Covariates included in this study included sex, race/ethnicity, family income, education, insurance status, region, cardiovascular risk factor (CRF) profile, and number of chronic comorbidities. Categorical variables were classified as follows: 2 categories for sex; 4 categories for race/ethnicity (non-Hispanic white, non-Hispanic black, non-Hispanic Asian, and Hispanic); 2 categories for family income (based on percent of family income relative to the federal poverty limit from the Census Bureau: middle/high income [≥200%] and low income [<200%]); 2 categories for education (some college or higher, high school/general equivalency diploma or less than high school); 2 categories for insurance type (insured and uninsured), and 4 categories for geographic region (Northeast, Midwest, South, and West). A CRF profile was derived from self-reported presence of each of the following risk factors: diagnosis of hypertension or high cholesterol, obesity (calculated body mass index ≥30 kg/m²), current smoker, or insufficient physical activity (based on not participating in >150 minutes per week of moderate-intensity aerobic physical activity, >75 minutes per week of vigorous-intensity aerobic physical activity, or a total combination of ≥150 minutes per week of moderate/vigorous-intensity aerobic physical activity). Based on the presence of these individual risk factors, an individual’s CRF profile was categorized as poor (≥4 CRFs), average (2–3 CRFs), and optimal (0–1 CRFs). Self-reported chronic comorbidities, including atherosclerotic cardiovascular disease (coronary heart disease, myocardial infarction, angina, and stroke), emphysema, chronic obstructive pulmonary disease, asthma, gastrointestinal ulcer, cancer (any), arthritis (including arthritis, gout, fibromyalgia, rheumatoid arthritis, and systemic lupus erythematosus), any kind of liver condition or weak/failing kidneys, were aggregated, and categorized as having 0, 1, or ≥2.²⁵

Statistical Analysis

χ² tests were used to compare general characteristics, and weighted proportions were used to study prevalence in our study population. We used unadjusted and adjusted logistic regression models to measure the association between measures of financial hardship from medical bills and the following explanatory variables: age (18–39 versus 40–64 years), sex, race or ethnicity, education, insurance status, family income, family size, region, years since diabetes mellitus diagnosis (<10 versus ≥10), CRF, and comorbidities. Furthermore, we assessed the association between measures of financial hardship from medical bills and their potential consequences: high financial distress, food insecurity, CRN, and foregone/delayed medical care. We obtained the variance estimation for the entire pooled cohort from the Integrated Public Use Microdata Series (http://www.ipums.org).²⁶ The percentage of missing values was <5% for all variables. For all statistical analyses, P<0.05 was considered statistically significant. All analyses were performed using Stata, version 15.1 (StataCorp, LP, College Station, TX). All analyses were survey-specific by using Stata’s svy functions, which additionally use person weights and variance estimates (based on person sampling units and stratum) to account for the complex survey design of NHIS to achieve nationally representative results. Since this study aggregated multiple years for analysis, all person weights were pooled and divided by the number of years studied, in accordance with NHIS guidance.¹³

Results

From 2013 to 2017, the NHIS total sample consisted of 164 696 surveyed individuals, of which 17 391 reported having diabetes mellitus (translated to 22.7 million adults annually, a prevalence of 9.6% [95% CI, 9.4–9.7]). A total of 123 706 were 18 to 64 years of age, of which 8967 reported having diabetes mellitus, representing 13.1 million nonelderly adults annually (prevalence of 6.8% [95% CI, 6.6–7.0]) across the United States (Figure I in the Data Supplement). The mean age was 51.6 years (SD 10.3) and 49.1% were female. Most nonelderly individuals with diabetes mellitus were insured non-Hispanic whites, from middle/high-income families, with an average/poor CRF profile (Table 1).

Financial Hardship Prevalence and Predictors

Annually, an estimated 5.4 million (41.1% [95% CI, 39.7–42.5]) nonelderly adults with diabetes mellitus were part of families that reported having financial hardship from medical bills, with 2 million (15.6% [95% CI, 14.6–16.6]) reporting an inability to pay medical bills at all.

Financial hardship from medical bills was more prevalent among those from low-income families, with lower education, lack of insurance, and a worse clinical profile (Table 2). In stratified analyses by family income and insurance, any financial hardship from medical bills was driven by lack of insurance, with almost equal prevalence between low-income and middle/high-income groups (Figure 1A). However, an inability to pay bills at all was significantly more prevalent among the low-income/uninsured group (39.1%), followed by middle/high-income/uninsured (23.2%), low-income/insured (21.6%), and middle/high-income/insured (8.5%; Figure 1B).

• Download figure
• Download PowerPoint

In multivariate analyses, after adjusting for known confounders, individuals who were uninsured had 2.26 (95% CI, 1.82–2.81) and 3.22 (95% CI, 2.61–3.97) higher odds of being in families with any financial hardship from medical bills and unable to pay bills at all, respectively. Similarly, those from low-income families had 1.21 (95% CI, 1.05–1.40) and 2.40 (1.96–2.94) higher odds of being in families with any financial hardship from medical bills and unable to pay bills at all, respectively. In a similar fashion, non-Hispanic blacks, those with a higher comorbidity count and a larger family size also had increased odds of financial hardship from medical bills and an inability to pay them. Older individuals (when compared with 18–39 years) had lower odds of suffering from either financial hardship from medical bills (odds ratio [OR], 0.74 [95% CI, 0.59–0.91]) or an inability to pay bills at all (OR, 0.58 [95% CI, 0.44–0.77]; Tables 3 and 4).

Financial Hardship Consequences

Among individuals with diabetes mellitus, those in families with financial hardship from medical bills had higher financial distress (52.1% versus 25.4%), food insecurity (30.0% versus 12.8%), CRN (34.7% versus 9.1%), and foregone/delayed medical care (55.5% versus 21.6%) when compared with those without it (all P<0.001). Moreover, there was a stepwise increase in the prevalence of high financial distress, food insecurity, CRN, and foregone/delayed medical care when comparing the graded categories of financial hardship from medical bills (Figure 2; Figure II in the Data Supplement). After adjusting for sociodemographic, economic, and clinical confounders, these associations remained significant (Table 5). Uninsured individuals from low-income families had the highest prevalence of each consequence, except for food insecurity, where the prevalence was similar between insured and uninsured individuals with low income (Figure III in the Data Supplement).When these consequences were measured in combination at a personal level, the likelihood of having ≥3 of them was 6-fold higher among those with an inability to pay bills at all compared with no financial hardship from medical bills (6% versus 38%, P<0.001; Figure 3). Furthermore, among those with an inability to pay bills at all, 18% reported having all 4 consequences.

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

Comparison With Nonelderly Individuals Without Diabetes Mellitus

When compared with those without diabetes mellitus, nonelderly individuals with diabetes mellitus had higher overall prevalence of financial hardship (41.1% versus 28.6%, P<0.001), as well as its consequences: high financial distress (35.6% versus 22.8%), food insecurity (19.9% versus 10.0%), CRN (19.8% versus 11.1%), and foregone/delayed medical care (35.5% versus 21.0%; P<0.001 for all). This association persisted after adjusting for known confounders: individuals with diabetes mellitus were at higher odds of experiencing financial hardship (OR, 1.27 [95% CI, 1.18–1.36]), high financial distress (OR, 1.14 [95% CI, 1.05–1.24]), food insecurity (OR, 1.27 [95% CI, 1.16–1.40]), CRN (OR, 1.43 [95% CI, 1.30–1.57]), and foregone/delayed medical care (OR, 1.30 [95% CI, 1.20–1.40; Figure IV in the Data Supplement). Overall, those with both diabetes mellitus and financial hardship reported were more likely to report these consequences (Figure 4).

• Download figure
• Download PowerPoint

Comparison With Elderly Individuals With Diabetes Mellitus

When compared with their elderly counterparts, nonelderly individuals with diabetes mellitus had twice the prevalence of financial hardship from medical bills (41.1% versus 20.7%, P<0.001, respectively; Table II in the Data Supplement). This pattern was found for the financial hardship consequences as well, with elderly individuals with diabetes mellitus having nearly half the prevalence for each consequence relative to nonelderly individuals with diabetes mellitus (Figure V in the Data Supplement). Overall, even when stratified by financial hardship status, elderly individuals had a significantly lower prevalence of zero financial hardship consequences when compared to nonelderly: 84% versus 42% among those with no financial hardship from medical bills, 62% versus 26% among those with financial hardship from medical bills but able to pay, and 46% versus 6% among those unable to pay bills at all (all P<0.001; Figure 3; Figure VI in the Data Supplement).

Discussion

Our study highlights that a substantial proportion of nonelderly patients with diabetes mellitus and their families in the United States struggle with medical bills and its deleterious consequences. We found that nearly 2 in 5 nonelderly individuals with diabetes mellitus in the US—or an estimated 5.4 million individuals annually—reported financial hardship from medical bills including problems affording medical care expenses within the past year or being burdened by medical debt that is payed over time. Notably, 1 in 6 persons with diabetes mellitus, or an estimated 2 million individuals annually, reported being in a family that was unable to pay bills at all. The financial burden from medical bills was especially substantial among those who were part of low-income families, lacked insurance, were non-Hispanic black, and had a higher comorbidity burden. This imposes not only a financial burden on these families but also limits their access to basic needs and adequate usage of healthcare services, with those with financial hardship reporting higher proportions of financial distress, food insecurity, CRN, and foregone/delayed medical care.

Financial burden from medical bills is substantial across the spectrum of patients with diabetes mellitus, irrespective of demographic, and socioeconomic status among those without healthcare insurance. In our report, 6 out of every 10 uninsured individuals with diabetes mellitus reported difficulty paying bills, with half of them (nearly 1 in 3) unable to pay at all. The likelihood of having the inability to pay medical bills at all for uninsured individuals was 2-fold higher than for those with insurance. Furthermore, low-income families without insurance were at the highest risk of having unaffordable bills. While among nonelderly adults with diabetes mellitus health insurance coverage increased from 84.7% in 2009 to 90.1% in 2016 among nonelderly adults with diabetes mellitus, our findings underscore the ongoing continuing need for attention by policymakers to expand further health insurance access for patients with diabetes mellitus, especially for low-income families.²⁷

Our study also highlights that while among patients with diabetes mellitus insurance coverage is critical to protect against risk of financial hardship from unexpected medical bills among patients with diabetes mellitus, current insurance structures fall short in this regard. In fact, the majority (83.2%) of patients with diabetes mellitus reporting problems paying bills are insured. Li et al have shown that 20% to 23% of families with diabetes mellitus with either private or public insurance spent a significant portion of their income on OOP medical expenses.¹¹ In our national sample of insured nonelderly adults with diabetes mellitus, nearly 1 in 7 (15.7%) reported inability to pay medical bills, which suggests inadequate protection (underinsurance) against substantial financial impact of OOP health expenses and with much higher reported burden among the low-income families. As a subanalysis, we also demonstrated that elderly individuals with diabetes mellitus have far lower rates of financial hardship from medical bills and its consequences when compared with nonelderly individuals with diabetes mellitus. However, at least 20% of elderly adults still suffer from financial hardship despite universal care. As we advance the debate beyond the expansion of health insurance coverage, policymakers should be concerned that current insurance arrangements do not adequately address the constraints of high health care costs and affordability. Although provisions in recent health care reforms set upper limits on OOP payments, these provisions do not adequately account for the financial challenges associated with high-risk conditions, such as diabetes mellitus,²⁸ especially for low-income families.²⁹ With the majority of the families in the United States with limited savings/assets to cover these unexpected expenses, the current trends toward shifting of a greater proportion of cost-sharing from payers to patients (underinsurance) will likely further exacerbate risk of financial hardships from medical care.

Our study further adds to the existing literature by highlighting potential tradeoffs associated with financial hardship from medical bills among diabetes mellitus patients. Nearly 1 in 6 diabetes mellitus individuals with financial hardship from medical bills reported either significant financial distress, cutting back on basic necessities, such as food and forgoing medical care. In fact, 1 in 2 of those unable to pay medical bills at all reported having at least 2 of these consequences, and 1 in 3 reported all 3. Of particular concern is the association of financial hardship from medical bills with medication nonadherence—a maladaptive coping strategy. While it may provide temporary economic relief to patients and families constrained with medical debt, medication nonadherence is associated with dire consequences for patients with diabetes mellitus, including worse glycemic control,³⁰ increased rates of health care utilization,^31–34 greater health care costs,^32–36 and higher mortality.^31,35

We also observed that the risks of these unintended consequences were consistently higher among patients with versus without diabetes mellitus. Since diabetes mellitus is associated with higher medical expenditures,^7,37 it is not surprising that patients with diabetes mellitus experience higher financial struggle than those without it. Patel et al¹⁹ described similar findings, showing that patients with diabetes mellitus had higher prevalence of food insecurity and CRN and were more prone to adopt cost-reducing behavioral strategies. We expand this knowledge by showing the stepwise contribution of both diabetes mellitus and financial hardship on these consequences; our results reveal that these associations remained significant even after adjusting for potential confounders (demographics, socioeconomic status, and comorbidities profile).

Our study findings illuminate the health and nonhealth related consequences as a potential side effect of burdensome medical bills from diabetes mellitus management has critical implications for policy, behavioral interventions, and clinical practice. Among others, these findings underscore need for price transparency and clear communication with patients and their families on costs of care. Great consideration should be taken on presenting less expensive but still effective treatment options, as well as need to comprehend patient’s insurance coverage and financial obligations in our management decision-making processes. Moreover, future interventions aimed at improving medication adherence should particularly focus on cost-related barriers and limiting financial toxicity from medications. Government regulations and public health policies should supplement these strategies by guaranteeing access to life-saving medications. Particularly, insulin total and OOP cost in the United States is on the rise and already higher than in any other high-income country, constituting one of the major barriers to treatment access beyond coverage.^38–40

Our study’s findings should be interpreted in light of the following limitations. First, our definition of financial hardship from medical bills is based on the survey data already available and other methods and questions could have more reliability in assessing it. However, this method has been used in prior studies and shown to correlate well with objective measures of financial burden from OOP costs for medical expenditures.⁴¹ Second, the questions used in NHIS assessed whether anyone in the household had financial hardship from medical bills in general, impeding us to provide a direct assessment of what proportion of the medical bills were related to diabetes mellitus. Third, we assessed presence/absence of diabetes mellitus based on the answers given by one randomly chosen adult per household. This, however, precluded us from knowing if other adults in the household had diabetes mellitus and would especially be important in those cases where the respondent answered “No” to having diabetes mellitus, but other adults in the household might have had it. However, we think that in doing so, the results from our study likely establish a more conservative approach to our calculations when comparing individuals with diabetes mellitus versus those without it, with any potential bias present being directed towards the null. Fourth, because of the cross-sectional design of the NHIS, we are precluded from establishing causality, and even after adjusting for all known and established covariates, residual confounding remains a possibility. Fifth, there might be participants overlap in the combined data from multiple years, and there are no publicly available identifiers to address this by excluding repeated representations in our study sample. However, the weighted estimate applies to US population and our methods follow the best practices laid out in the accompanying documentation of the NHIS.

In conclusion, we found that nonelderly patients with diabetes mellitus have a high prevalence of financial hardship from medical bills, which is associated with high financial distress, food insecurity, CRN, and foregone/delayed medical care, therefore, increasing the probability of higher medical and nonmedical unmet needs.

Footnotes