To examine the relationship between chronic health conditions and out-of-pocket costs (OOPC) and medical debt.
Secondary data from the 2013 Panel Study of Income Dynamics (PSID) was used. Households whose head of household and spouse (for married households) were 18 to 64 years old were included.
Households with 1 to 3 chronic conditions had higher odds of having any OOPC compared to households with no chronic conditions (AOR 1.74, 95% CI 1.39, 2.17) (p < .01). Households with 1 to 3 and 4 or more chronic health conditions were associated with higher odds of having any medical debt (AOR 2.24, 95% CI 1.75 to 2.87; AOR 5.04, 95% CI 3.04 to 8.34) compared to those with no chronic conditions (p < 0.01). Similarly, 1 to 3 and 4 or more chronic health conditions was associated with higher amounts of OOPC (Exponentiated Coefficient 1.18, 95% CI 1.03 to 1.36; Exponentiated Coefficient 1.56, 95% CI 1.17 to 2.07) and medical debt (Exponentiated Coefficient 1.69, 95% CI 1.23 to 2.34; Exponentiated Coefficient 2.73, 95% CI 1.19 to 6.25) compared to households with no chronic conditions (p < 0.05).
Citation: Richard P, Walker R, Alexandre P (2018) The burden of out of pocket costs and medical debt faced by households with chronic health conditions in the United States. PLoS ONE 13(6): e0199598. https://doi.org/10.1371/journal.pone.0199598
Editor: Gianni Virgili, Universita degli Studi di Firenze, ITALY
Received: July 24, 2017; Accepted: June 11, 2018; Published: June 25, 2018
This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Data Availability: All 2013 Panel Study of Income Dynamic (PSID) files are available from the PSID Public Data Extract Repository (http://doi.org/10.3886/E100784V1).
Funding: This study was funded by the USU/PMB Incoming Faculty Research Support Start-up Package (Grant # C087328915). Funding was received by P.R. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The rise in treatment costs and associated cost sharing for individuals and families with chronic health conditions have been found to impose a significant financial burden on families or individuals, even those with health insurance. The financial burden of adverse health events ranges from filing for bankruptcies [1–3] (the extent of which is controversial, for e.g., [1,2,3]) to medical debt [4–8]. Furthermore, uninsured families with medical debt have even depleted their savings and other assets or forgone the consumption of necessities such as food, heat or rent to pay for their medical debt [6, 7].
More specific to our study, there is an emerging literature on the relationship between chronic health conditions and consumer debt in the elderly using data from the Health and Retirement Study (HRS) [9, 10]. While these studies have provided some insight regarding the impact of chronic health conditions on consumer debt, they tend to focus mostly on the elderly population. One exception is the study by Dobkin et al.  that examined the impact of hospital admissions on out of pocket spending and unpaid bills for non-elderly adults aged from 50 to 64 years. But, none of these studies focus on specific health adverse events such as chronic health events. Our study investigates the association between chronic health conditions and measures of financial burden in the adult population from 18 to 64 years old. The goal of the current study is to examine the relationship between chronic health conditions and financial burden (out of pocket costs and medical debt) outcomes in the non-elderly population.
This study hypothesized that chronic health conditions would be associated with medical debt outcomes through two major pathways. The first major pathway is through large and repeated out-of-pocket costs (OOPC) of treatment. There is an extensive literature, mostly in the elderly population, that has documented the effect of chronic health conditions on OOPC using data from the HRS or Medicare claims files [11–20]. Similar results have been found among the non-elderly, where about 13% of individuals incurred OOPC exceeding 20% of their annual income .
The second major pathway is through the loss of employment. Households that experience chronic health conditions may also simultaneously experience a “productivity” shock . On the one hand, chronic health conditions make it difficult for patients to balance activities such as employment while seeking medical treatment, thereby increasing their taste for leisure. On the other hand, chronic health conditions might encourage labor force participation for those who are able to work because of the link between employment and health insurance in the United States . Most studies in this area tend to focus on the impact of cancer treatment and survivorship on the labor markets’ outcomes of patients . Other chronic conditions such as diabetes and depression have also been found to negatively impact an individual’s ability to participate in the labor markets and can result in a loss of income or reduction in wages [24, 25].
More importantly, as health insurance plays a key role in the relationship between chronic health conditions and medical debt outcomes, it is essential to consider some emerging but well-designed studies in this area that seek to understand to what extent that providing coverage to uninsured patients reduces the financial burden associated with treatment. In this case, four recent studies using data from the Massachusetts Health Insurance Reform of 2006 and the Oregon Health Insurance Experiment of 2008 have found that insurance coverage reduced the share and total amount of debt that was past due, including medical debt for newly insured residents [26–29].
There is limited understanding in the literature on the relationship between chronic health conditions and OOPC and medical debt outcomes at the household level. It is important that decision makers and policy makers are informed of and understand this relationship because of the high financial burden that these conditions might impose on indivdiduals, households and society at large.
Data source and subjects
Data from the 2013 Panel Study of Income Dynamics (PSID) was used. The PSID is a nationally representative panel survey that collects socioeconomic and demographic data on households and individuals in the United States. One round of interviews is conducted per panel. The 2013 PSID includes 9,063 households. This study uses households as the unit of analysis. Analyses were done at the household level because the dependent variables are measured at this level. Households whose head of household and spouse (for married households) were 18 to 64 years old were included (see Fig 1). Only households that formed a “health insurance eligible unit” were included (i.e., those with non-family members such as relatives and friends living in the home were excluded). One reason for this is because the head of household is unlikely to incur debt on their behalf. Moreover, including them might bias the results as some of these families might have chronic health conditions of their own. A further exclusion included households whose head or spouse if married was 65 years old or older because of the study’s focus on the non-elderly. “Dual eligibles,” beneficiaries of both Medicare and Medicaid, were also excluded because of unique health needs that may interact with chronic health conditions. Most of the households that were not included were based on these criteria; the remaining, 310 households, had missing observations (given the small number of missing data, we do not believe that it would have an impact on the results). Consistent with the literature in this area, for outcomes of the amount of OOPC and medical debt, only households with positive amounts were included [3, 30–32]. The analytic sample for the outcome of amount of OOPC was 3,882 of the 6,042 households with any OOPC and amount of medical debt was 664 of the 6,028 households with any medical debt.
The figure depicts the number of households in the 2013 PSID followed by the criteria used for exclusion and the number of households excluded, and then the final sample sizes for households with any out-of-pocket costs (OOPC), any medical debt, positive amounts of OOPC and positive amounts of medical debt.
Dependent variables included: 1) any OOPC, 2) amount of OOPC (OOPC > 0; $2013); 3) any medical debt and; 4) amount of medical debt (medical debt >0; $2013). OOPC was determined by the responses to the following questions: “About how much did (you (and your family)/they) pay out-of-pocket for nursing home and hospital bills in 2011 and 2012 combined?” and “About how much did you (and your family) pay out-of-pocket for doctor, outpatient surgery, and dental bills in 2011 and 2012 combined?” Medical debt was determined by the response to the following: “If you added up all medical bills [respondents/family members living there], about how much would they amount to right now? INCLUDE unpaid balance(s), or medical bills that are outstanding.” Responses are validated by using bills and other documents from the respondents. Several studies have found these measures to be highly reliable [33, 34]. Only those with outstanding medical bills were considered. The National Health Interview Survey measures medical debt similarly [8, 35]. A binary measure was created for any OOPC and any medical debt being coded as 1 if the respondent has any OOPC or medical debt, respectively, or 0 otherwise. Amount of OOPC and amount of medical debt were continuous variables reflecting the amount of OOPC and medical debt, respectively, as indicated by the respondent. Missing values were imputed by the PSID.
The key independent variable for this study is chronic conditions that included hypertension or high blood pressure, heart disease, diabetes, asthma, lung disease, cancer, schizophrenia, bipolar disorder, stroke, heart attack, and arthritis was created . For each chronic condition, respondents were asked “Has a doctor or other health professional EVER told (you/HEAD) that (you/he/she) had …CONDITION?”. Based on the distribution of chronic conditions in the study sample, the number of self-reported chronic health conditions was categorized into no chronic conditions, 1 to 3 chronic conditions, and 4 or more chronic conditions. Variables associated with consumer debt or measures of financial burden outcomes were controlled for [10, 26, 28, 29, 37]. The head of household’s gender, age category, race, marital status, and education and household characteristics (both the head and spouse if married) of insurance coverage status, number of chronic health conditions, and number of children were controlled for. Geographic region variables were included to capture differences in health insurance and medical debt outcomes by location. The amount of OOPC, amount of medical debt and number of children were log transformed to adjust for the skewness in the distribution of the variables. In order to account for an oversampling of immigrant and minority households and for attrition, family weights were utilized .
Logistic regression models that adjusted for a set of covariates were used to estimate the relationship between chronic health conditions and the likelihood of having any OOPC or medical debt. Odds ratios were presented for these models for better interpretation of the results. Generalized linear models (GLMs) with logarithmic link and a gamma distribution were used to estimate the relationship between chronic health conditions and the amount of OOPC (> 0) or medical debt (> 0) (see ). Exponentiated coefficients were presented for these models for better interpretation of the results.
The total number of households for the study is 5,860 and 78% of households had any type of debt. For the outcomes of the study, 73% of the households had any OOPC and 9% had any medical debt. The average amount of OOPC was $1171 and the average amount of medical debt was $838.
In terms of analyses, we examined the relationship between chronic health conditions and financial outcomes by categorizing chronic conditions into no chronic conditions, 1 to 3 chronic conditions and 4 or more chronic conditions. For households that had no chronic conditions, 68% had any OOPC, while for those with 1 to 3 chronic conditions, 80% had any OOPC, and for those with 4 or more chronic conditions, 78% had any OOPC (Table 1). These differences were statistically significant between households with no chronic conditions and 1 to 3 chronic conditions (p < .01) and households with no chronic conditions and 4 or more chronic conditions (p < .05). For amount of OOPC, we found $968.25 for households with no chronic conditions, $1362.94 for households with 1 to 3 chronic conditions, and $1950.91 for households with 4 or more chronic conditions. These differences were statistically significant between households with no chronic conditions and 1 to 3 chronic conditions (p < .01) and households with no chronic conditions and 4 or more chronic conditions (p < .01). For households that had no chronic conditions, 7% had any medical debt, while for those with 1 to 3 chronic conditions, 12% had any medical debt, and for those with 4 or more chronic conditions, 19% had any medical debt. These differences were statistically significant between households with no chronic conditions and 1 to 3 chronic conditions (p < .01) and households with no chronic conditions and 4 or more chronic conditions (p < .01). For amount of medical debt, we found $359.48 for households with no chronic conditions, $1,267.06 for households with 1 to 3 chronic conditions, and $2986.68 for households with 4 or more chronic conditions. These differences were statistically significant between households with no chronic conditions and 1 to 3 chronic conditions (p < .01) and households with no chronic conditions and 4 or more chronic conditions (p < .05).
Households with 1 to 3 chronic conditions compared to households with no chronic conditions had 1.74 odds of having any OOPC (95% CI 1.39 to 2.17) (p < .01) (Table 2). However, beyond 1 to 3 chronic conditions, there was a non-linear association between households and chronic conditions and the odds of having any OOPC. For instance, while households with 4 or more chronic conditions have higher odds of having any OOPC compared to households with no chronic conditions, the results were not statistically significant. Adjusted odds ratios showed that households in which members (the head and spouse) had 1 to 3 and 4 or more chronic health conditions were associated with higher odds of having any medical debt (AOR 2.24, 95% CI 1.75 to 2.87; AOR 5.04, 95% CI 3.04 to 8.34) compared to those households where members had no chronic conditions (p < 0.01). Similarly, exponentiated coefficients from GLM models that used a log link with gamma distribution for households with positive amounts of OOPC and medical debt found that 1 to 3 and 4 or more chronic health conditions were associated with higher amounts of OOPC (Exponentiated Coefficient 1.18, 95% CI 1.03 to 1.36; Exponentiated Coefficient 1.56, 95% CI 1.17 to 2.07) and the amount of medical debt (Exponentiated Coefficient 1.69, 95% CI 1.23 to 2.34; Exponentiated Coefficient 2.73, 95% CI 1.19 to 6.25) compared to households with no chronic conditions (p < 0.05). These represent increases over the constant terms of about $174 and $542 in the amount of OOPC for 1 to 3 and 4 or more chronic conditions, respectively and $248 and $622 in the amount of medical debt for 1 to 3 and 4 or more chronic conditions, respectively.
Additionally, households that were headed by females were more likely to have any OOPC or any medical debt compared to households that were headed by males. Interestingly, households headed by those who were between 45 and 64 years old were more likely to have any OOPC but less likely to have any medical debt compared to those where the head was between 18 and 34 years old. Medicaid beneficiary households were less likely to have any OOPC compared to those that were privately insured. Uninsured households were less likely to have any OOPC compared to privately insured households. However, for those with any OOPC, uninsured households were more likely to have higher amounts of OOPC compared to those with private health insurance. These findings are consistent with similar studies in the literature [40, 41], which reflect issues of access to health care by those who do not have health insurance, but experience higher amounts of OOPC when accessing health services compared to those who are privately insured. Those who live in the South and the Midwest were more likely to have any medical debt compared to those who live in the northeastern region of the United States. These may reflect the variations in health insurance across the different regions of the United States. In terms of amount, households headed by 45 to 64 year olds were associated with higher amounts of OOPC compared to those headed by individuals who were between 18 and 34 years old. Similarly, households that lacked health insurance or headed by Medicaid beneficiaries were associated with a higher amount of medical debt compared to those with private insurance.
Findings showed that 1 to 3 chronic health conditions were associated with higher odds of having any OOPC and 1 to 3 and 4 or more chronic health conditions were associated with higher odds of having any medical debt at the household level. Similarly, for households with positive OOPC and medical debt, 1 to 3 and 4 or more chronic health conditions were associated with higher amounts of OOPC and medical debt, representing increases over the constant terms of about $174 and $542 in the amount of OOPC for 1 to 3 and 4 or more chronic conditions, respectively and $248 and $622 in the amount of medical debt for 1 to 3 and 4 or more chronic conditions, respectively. Findings from our study are consistent with those of the studies conducted by Kim et al. (2012) and Barbiarz et al. (2013) [9, 10] that have looked at the association between adverse health events and unsecured debt using data from the HRS. Defining a new health event as those households where the head has been diagnosed with any of the following conditions including high blood pressure, diabetes, cancer, lung disease, heart problems, stroke, arthritis, or psychological problems, Barbiarz et al. (2013) [9, 10] found that new health events are associated with a 10% increase in the probability of having any unsecured debt and approximately an 11% increase in the amount of unsecured debt. Converting the odds ratios from the logistic regressions into marginal effects by using the marginal effects command from STATA 14 (results not presented) and the marginal effects for regression coefficients for dummy variables calculated as ex − 1 (see Table 2) [3, 42], we found that a change by household members from having no chronic conditions to having 1 to 3 chronic conditions was associated with a change in the probability of having any OOPC by 15% and any medical debt by 84%. Further, a change by household members from having no chronic conditions to having 4 or more chronic conditions was associated with a change in the probability of having any medical debt by 168%. However, the exponentiated coefficients from Table 2 (columns 2 and 4) showed that 1 to 3 and 4 or more chronic health conditions were associated with increased amounts of OOPC by 18% and 56% and medical debt by 69% and 173% for those households with positive amounts of OOPC and medical debt. These results are much greater in magnitude compared to those found by Kim et al. (2012), Barbiarz et al. (2013), and Dobkin et al. [9, 10]. There are several plausible explanations for these differences in the amounts of OOPC and medical debt including the fact that our sample has a higher proportion of uninsured households, examines medical debt as opposed to unsecured debt and considers OOPC for patients in inpatient hospital services whether the head or spouse experiences the chronic health condition in the household. Additionally, contrary to Dobkin et al.  whose study used a sample of patients admitted to hospitals in California, our study used broader measures of OOPC and medical debt that included nursing homes, hospital bills, ambulatory care and dental care. It is important to examine the relationship between chronic conditions and measures of financial burden for insured and uninsured households aged 18 and 64 years old to understand how these conditions might impact different age groups and households by types of insurance status differently.
There are some limitations that are worth considering in this study. Although a high proportion of the medical debt amount was validated by the PSID through verification of the actual medical bills , the chronic health conditions were self-reported. Further, it is also important to note that different types of chronic conditions such as cancer or stroke may have a differential association with OOPC and medical debt outcomes compared to other illnesses such as bipolar disorder or asthma. Along similar lines, the severity of the illness and details on the management of the illness were not known.
The findings from this study can be explored in the context of public policy, such as in the Affordable Care Act (ACA). State and federal policy changes to decrease the financial burden of households with chronic health conditions have recently occurred . Nonetheless, findings from the current study show that chronic health conditions still have a financial burden on some households. Additionally, the study results can inform medical professionals about the financial impact of clinical decisions and treatments for patients with chronic health conditions.
The essential benefits coverage provisions of health services from the ACA have important implications for households whose chronic health conditions are associated with medical debt or OOPC. For example, certain preventive services must be included without cost-sharing. An increase in preventive services could decrease inpatient hospital services for chronic health conditions while decreasing the household’s financial burden. Further research might want to consider the impact of the ACA on medical debt and OOPC for households where members have experienced chronic health conditions.
Findings showed that 1 to 3 chronic health conditions were associated with higher odds of having any OOPC and 1 to 3 and 4 or more chronic health conditions were associated with higher odds of having any medical debt. For households with positive amounts of OOPC and medical debt, 1 to 3 and 4 or more chronic health conditions were associated with higher amounts of OOPC and medical debt. These findings showed that the presence of chronic health conditions impose a large financial burden on some households. An important extension of this paper would be to use instrumental variables to account for any potential omitted variable bias or reverse causality. Future research may also want to look at longitudinal data to assess the impact of chronic conditions on medical debt and OOPC.
No individuals contributed to this work who do not meet PLOS ONE’s authorship criteria.
Disclaimer: The views expressed are those of the author and do not necessarily reflect the official views of the Uniformed Services University of the Health Sciences, the Henry M. Jackson Foundation for the Advancement of Military Medicine or the Department of Defense.
- 1. Dobkin A, Finkelstein, A., Kluender, R., and M. Notowidigdo. The Economic Consequences of Hospital Admissions. National Bureau of Economic Research Working Paper 22288. 2015.
- 2. Himmelstein DU, Thorne D, Warren E, Woolhandler S. Medical Bankruptcy in the United States, 2007: Results of a National Study. The American Journal of Medicine. 2009;122(8):741–6. pmid:19501347
- 3. Cameron AC, Trivedi PK. Microeconometrics using stata: Stata press College Station, TX; 2008.
- 4. Cunningham P. Trade-Offs Getting Tougher: Problems Paying Medical Bills Increase for U.S. Families, 2003–2007. Center for Studying Health System Change Tracking Report 2008(21).
- 5. Doty M, S. Collins S. Rustgi, and Kriss J. Seeing Red: The Growing Burden of Medical Bills and Debt Faced by U.S. Families. The Commonwealth Fund Issue Brief. 2008.
- 6. Dranove D, Millenson ML. Medical bankruptcy: Myth versus fact. Health Affairs. 2006;25(2):w74–w83. pmid:16507555
- 7. Gross T, Notowidigdo MJ. Health insurance and the consumer bankruptcy decision: Evidence from expansions of Medicaid. Journal of Public Economics. 2011;95(7–8):767–78.
- 8. Pollitz K, Cox C, Lucia K, Keith K. Medical debt among people with health insurance. Kaiser Family Foundation, 2014.
- 9. Kim H, Yoon W, Zurlo KA. Health shocks, out-of-pocket medical expenses and consumer debt among middle-aged and older Americans. Journal of Consumer Affairs. 2012;46(3):357–80.
- 10. Babiarz P, Widdows R, Yilmazer T. Borrowing to cope with adverse health events: Liquidity constraints, insurance coverage, and unsecured debt. Health Economics. 2013;22(10):1177–98. pmid:23070846
- 11. Webb A, Zhivan NA. How Much is Enough? The Distribution of Lifetime Health Care Costs: Center for Retirement Research at Boston College; 2010.
- 12. De Nardi M, French E, Jones JB. Why do the elderly save? The role of medical expenses. National Bureau of Economic Research, 2009.
- 13. Paez KA, Zhao L, Hwang W. Rising out-of-pocket spending for chronic conditions: A ten-year trend. Health affairs. 2009;28(1):15–25. pmid:19124848
- 14. Meraya AM, Raval AD, Sambamoorthi U. Chronic Condition Combinations and Health Care Expenditures and Out-of-Pocket Spending Burden Among Adults, Medical Expenditure Panel Survey, 2009 and 2011. Preventing Chronic Disease. 2015;12:E12. pmid:25633487
- 15. Tu HT, Cohen GR. Financial and health burdens of chronic conditions grow. Center for Studying Health System Change Tracking Report. 2009;24:1–6.
- 16. Bestvina CM, Zullig LL, Yousuf Zafar S. The Implications of Out-of-pocket Cost of Cancer Treatment in the USA: A Critical Appraisal of the Literature. Future Oncology. 2014;10(14):2189–99. pmid:25471033
- 17. Davidoff AJ, Erten M, Shaffer T, Shoemaker JS, Zuckerman IH, Pandya N, et al. Out-of-pocket health care expenditure burden for Medicare beneficiaries with cancer. Cancer. 2013;119(6):1257–65. pmid:23225522
- 18. Kircher SM, Johansen ME, Nimeiri HS, Richardson CR, Davis MM. Impact of Medicare Part D on out-of-pocket drug costs and medical use for patients with cancer. Cancer. 2014;120(21):3378–84. pmid:24962682
- 19. Zafar SY, Peppercorn JM, Schrag D, Taylor DH, Goetzinger AM, Zhong X, et al. The financial toxicity of cancer treatment: A pilot study assessing out-of-pocket expenses and the insured cancer patient's experience. The Oncologist. 2013;18(4):381–90. pmid:23442307
- 20. Jaspers L, Colpani V, Chaker L, van der Lee SJ, Muka T, Imo D, et al. The global impact of non-communicable diseases on households and impoverishment: A systematic review. European Journal of Epidemiology. 2015;30(3):163–88. pmid:25527371
- 21. Bernard DS, Farr SL, Fang Z. National Estimates of Out-of-pocket Health Care Expenditure Burdens among Nonelderly Adults with Cancer: 2001 to 2008. Journal of Clinical Oncology. 2011;29(20):2821–6. pmid:21632508
- 22. Bradley CJ, Neumark D, Barkowski S. Does Employer-provided Health Insurance Constrain Labor Supply Adjustments to Health Shocks? New Evidence on Women Diagnosed with Breast Cancer. Journal of Health Economics. 2013;32(5):833–49. pmid:23891911
- 23. Moran JR, Short PF, Hollenbeak CS. Long-Term Employment Effects of Surviving Cancer. Journal of Health Economics. 2011;30(3):505–14. pmid:21429606
- 24. Chatterji P, Alegria M, Takeuchi D. Psychiatric Disorders and Labor Market Outcomes: Evidence from the National Comorbidity Survey-Replication. Journal of Health Economics. 2011;30(5):858–68. pmid:21782264
- 25. Ojeda VD, Frank RG, McGuire TG, Gilmer TP. Mental illness, nativity, gender and labor supply. Health Economics. 2010;19(4):396–421. pmid:19370696
- 26. Mazumder B, Miller S. The effects of the Massachusetts Health Reform on financial distress. Working Paper, Federal Reserve Bank of Chicago, 2014.
- 27. Baicker K, Finkelstein A. The Effects of Medicaid Coverage—Learning from the Oregon Experiment. The New England Journal of Medicine. 2011;365(8):683–5. pmid:21774703
- 28. Baicker K, Taubman SL, Allen HL, Bernstein M, Gruber JH, Newhouse JP, et al. The Oregon experiment—effects of Medicaid on clinical outcomes. New England Journal of Medicine. 2013;368(18):1713–22. pmid:23635051
- 29. Finkelstein A, Taubman S, Wright B, Bernstein M, Gruber J, Newhouse JP, et al. The Oregon health insurance experiment: Evidence from the first year. Quarterly Journal of Economics. 2012;127(3):1057–106. pmid:23293397
- 30. Manning WG. The logged dependent variable, heteroscedasticity, and the retransformation problem. Journal of Health Economics. 1998;17(3):283–95. pmid:10180919
- 31. Manning WG, Mullahy J. Estimating log models: To transform or not to transform? Journal of Health Economics. 2001;20(4):461–94. pmid:11469231
- 32. Mullahy J. Much ado about two: Reconsidering retransformation and the two-part model in health econometrics. Journal of Health Economics. 1998;17(3):247–81. pmid:10180918
- 33. Becketti S, Gould W, Lillard L, Welch F. The panel study of income dynamics after fourteen years: An evaluation. Journal of Labor Economics. 1988;6(4):472–92.
- 34. Fitzgerald J, Gottschalk P, Moffitt RA. An analysis of sample attrition in panel data: The Michigan Panel Study of Income Dynamics. National Bureau of Economic Research Cambridge, Mass., USA; 1998.
- 35. Cohen RA, Gindi RM, Kirzinger WK. Financial Burden of Medical Care: Early Release of Estimates from the National Health Interview Survey, January-June 2011. National Center for Health Statistics; 2012.
- 36. Sambamoorthi U, Tan X, Deb A. Multiple chronic conditions and healthcare costs among adults. Expert Review of Pharmacoeconomics & Outcomes Research. 2015;15(5):823–32.
- 37. Crook J, Hochguertel S. Household debt and credit constraints: Evidence from OECD countries. Credit Research Center, University of Edinburgh Working Paper Series; 2005.
- 38. PSID. PSID main interview user manual: Release 2015. Institute for Social Research, University of Michigan, 2015 June, 2015. Report No.
- 39. Richard P, Shin P, Beeson T, Burke LS, Wood SF, Rosenbaum S. Quality and Cost of Diabetes Mellitus Care in Community Health Centers in the United States. PloS one. 2015;10(12):e0144075. pmid:26636324
- 40. Halpern MT, Renaud JM, Vickrey BG. Impact of insurance status on access to care and out-of-pocket costs for US individuals with epilepsy. Epilepsy & Behavior. 2011;22(3):483–9.
- 41. Roemer M. Statistical Brief 500: Out-of-Pocket Health Care Expenses for Medical Services, by Insurance Coverage, 2000–2014. Rockville, MD: Agency for Healthcare Research and Quality. 2017.
- 42. Wooldridge JM. Econometric Analysis of Cross Section and Panel Data. Cambridge, MA: The MIT Press; 2002.
- 43. Beronio K, Glied S, Frank R. How the Affordable Care Act and Mental Health Parity and Addiction Equity Act greatly expand coverage of behavioral health care. The Journal of Behavioral Health Services & Research. 2014;41(4):410–28.