You are here

Article Text

Article menu

Download PDFPDF

Research report
Consumer credit, chronic disease and risk behaviours
  1. Lorraine T Dean1,2,
  2. Emily A Knapp1,
  3. Sevly Snguon1,
  4. Yusuf Ransome3,
  5. Dima M Qato4,
  6. Kala Visvanathan1,2
  1. 1 Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Johns Hopkins University, Baltimore, Maryland, USA
  2. 2 Department of Oncology, Johns Hopkins School of Medicine, Baltimore, Maryland, USA
  3. 3 Department of Social and Behavioral Sciences, Yale University School of Public Health, New Haven, Connecticut, USA
  4. 4 Department of Pharmacy Systems, Outcomes and Policy, University of Illinois College of Pharmacy, Chicago, Illinois, USA
  1. Correspondence to Dr Lorraine T Dean, Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD 21205, USA; lori.dean{at}jhu.edu

Abstract

Background Credit scores have been identified as a marker of disease burden. This study investigated credit scores’ association with chronic diseases and health behaviours that are associated with chronic diseases.

Methods This cross-sectional analysis included data on 2083 residents of Philadelphia, Pennsylvania, USA in 2015. Nine-digit ZIP code level FICO credit scores were appended to individual self-reported chronic diseases (obesity, diabetes, hypertension) and related health behaviours (smoking, exercise, and salt intake and medication adherence among those with hypertension). Models adjusted for individual-level and area-level demographics and retail pharmacy accessibility.

Results Median ZIP code credit score was 665 (SD=58). In adjusted models, each 50-point increase in ZIP code credit score was significantly associated with: 8% lower chronic disease risk; 6% lower overweight/obesity risk, 19% lower diabetes risk; 9% lower hypertension risk and 14% lower smoking risk. Other health behaviours were not significantly associated. Compared with high prime credit, subprime credit score was significantly associated with a 15%–70% increased risk of chronic disease, following a dose–response pattern with a prime rating.

Conclusion Lower area level credit scores may be associated with greater chronic disease prevalence but not necessarily with related health behaviours. Area-level consumer credit may make a novel contribution to identifying chronic disease patterns.

  • consumer credit
  • chronic disease
  • obesity
  • diabetes
  • hypertension
  • United States

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/jech-2018-211160

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Background

    A growing body of research has associated area-level and individual-level consumer credit with a number of health outcomes,1–7 with most focusing on general health and acute conditions. No previous studies have explored the association between area-level consumer credit, chronic diseases and behaviours associated with those diseases.

    Consumer credit scores are quantitative measures of a person’s financial history and cumulative financial decisions, based on the use and timely payment of loans, credit cards and debts.1 2 Credit scores trend higher with increasing income2 8 and older age, but there are gaps in knowledge about credit scores among those with low educational attainment.9 Scores may be lower among racial/ethnic minorities and lower among those in economically distressed communities, who are most likely to be targets of lower value and subprime credit products that are either not included or are penalised in scoring models.10 11 While these trends are seen in credit scores for individuals, individual scores are related to area-level scores, which we define as the aggregate mean of individual’s scores who reside in a geographic unit.

    Area-level scores represent averages of individual scores, such that a community with many low individual scores would have lower area-level scores or a community with many high individual scores would have higher area-level scores. One study showed that residents of areas recovering from a local economic downturn11 were assigned lower credit ratings, than those in economically stronger areas who had the same credit history. In this way, the local credit economy forms a foundation for individual credit scores, which may then reflect back onto an area’s overall future creditworthiness. These area-level scores characterise the local credit economy and are used by retail businesses to determine which products, services and interest rates will be advertised to individuals in certain areas.10 This may influence, for example, the location of retail pharmacies in that area.12 Previous work supports that retail pharmacies, particularly chains, are less likely to be located in socioeconomically challenged areas13 or medically underserved areas.14 Thus, both economic and demographic factors may be important to consider in investigations of credit scores and health, especially given that age and socioeconomic gradient also influence health outcomes.15–18

    An area-level credit score may be a measure of area-level socioeconomic position (SEP).5 SEP is an aggregate latent construct that includes both resource-based (income, wealth, education) and prestige-based (education, social connections, and status) measures that represent one’s social position and access to material goods.19 Credit scores reflect both of these properties and may have advantages over commonly used SEP measures, such as income and education. For example, while area-level household averages of income and education are commonly collected SEP measures, the nature of employment and assistance makes income challenging to capture among people who do not earn a regular salary, is often misreported or unreported and even when reported may not truly capture one’s actual economic circumstances and resources.20–22 According to the Federal Reserve, 20% of households reported outspending what they had earned in the past 12 months (May 2015 data8) indicating that income may not fully capture a household’s economic situation. Thus, education is more likely to be accurate and easy to assess in self-reports, but the economic and social returns of educational attainment vary by societal culture and age cohort.22 In comparison, consumer credit ratings and credit scores represent a single measure that can be readily obtained, that captures both resource-based and prestige-based SEP dimensions, and can capture both what economic opportunities are extended to members of a community and the economic assets present in a community; however, few studies have linked consumer credit to health conditions.1–7 23–31

    In the USA, obesity, hypertension and diabetes are among the top 10 chronic medical conditions for which patients receive treatments that incur high ongoing out-of-pocket costs,17 which may have direct implications for consumer credit. Smoking and lack of physical activity may elevate risk for these diseases.18 This study assessed area-level consumer credit scores and risk behaviours associated with the chronic diseases of obesity, diabetes and hypertension, adjusting for individual-level and area-level socioeconomic and demographic factors.

    Methods

    This cross-sectional study used de-identified data from a 2015 random digit dialling survey of Philadelphia, Pennsylvania residents ≥18 years old (Public Health Management Corporation Community Health Data Base, 2015 Southeastern Pennsylvania Household Health Survey). The response rates of 9.6% for landlines and 12.6% for cell phones fall within the range of other respected population-based health surveys (eg, California Health Interview Survey and Pew Internet and American Life Project). Individual-level survey responses were matched to US Census-derived 9-digit ZIP code demographic data calculated by Easy Analytic Software (Easy Analytic Software I. ZIP4 EASI Demographic Files. 2015). Retail pharmacy counts were based on data from the National Council for Prescription Drug Programs (NCPDP), which includes all active retail pharmacies as of February 2015. The study was deemed exempt from Institutional Review Board review.

    Outcomes

    The primary outcomes were self-reported: overweight or obese (BMI >25); prior diagnosis of diabetes, or hypertension; and among those with hypertension, currently watching or reducing sodium intake, and compliance with prescribed hypertension medication (adheres all of the time vs less than all of the time), current smoker and exercising ≥90 min per week.

    Exposure

    The exposure was Equifax average household FICO credit scores in 2015 in Philadelphia County, aggregated to 9-digit ZIP codes (n=2002) and merged to respondents based on their reported address of residence in 2015. FICO scores were calculated for a 9-digit zip code area when there were data from at least seven households. The smallest area unit available was the 9-digit ZIP code, which we used as a proxy for the local credit economy. FICO scores range from 300 to 850 where higher values represent better credit.

    Covariates

    Individual-level covariates included: self-reported age, race, ethnicity, annual household income, education, below 200% poverty and health insurance status. Area-level covariates were chosen as corollaries of individual-level covariates to account for the compositional contributions of neighbourhood demographics to credit scores and health. Area-level covariates included: median age, % non-Hispanic (NH) Black, % Hispanic/Latinx ethnicity, % with annual household income >$100 000 and number of retail pharmacies in Philadelphia County. Nine-digit ZIP code level variables were calculated by EASI Analytic Software. Comparable area-level poverty and health insurance prevalence estimates were not available.

    Analysis

    Frequencies and means were calculated for each covariate. The income variable was missing for 22% of respondents and was imputed using a logit model based on individual and area-level demographics. Spearman correlations between all variables were assessed to account for the non-normal distributions across 9-digit ZIP codes. Generalised estimating equation (GEE) models were used to obtain robust standard errors to account for potential clustering across ZIP codes. In the first set of analyses, continuous credit scores were divided into 50-point categories, such that the reported effect sizes are interpreted in terms of 50 point increments. The second set of analyses used the Federal Reserve designations of subprime (≤659; worst credit), near prime (660–719) and higher than prime (≥720, best credit).32 All models adjusted for area-level and individual-level income, education, age, race, ethnicity; individual-level insurance status and poverty; and area-level retail pharmacy counts. We also ran models with and without credit scores to check for attenuation of the relationship between race or racial composition and health outcomes. Outcomes were estimated on the relative risk scale. Statistical significance was assessed at p<0.05. Analyses were conducted using Stata V.14.

    Results

    We analysed data among 2083 respondents. The median are-level credit score was 665 (SD=58), with a range of 531–804, which was comparable to the median national credit score of 675 in 2015.33 Prevalence of behavioural risk factors and chronic disease outcomes were comparable to or higher than USA-based averages in 2015 for the population above 18 years.34 Prevalence of overweight/obese was 71%, diabetes was 17%, hypertension was 43%, smoking was 19% and regular exercise was 48%. Of those with hypertension, 81% were watching salt intake and 13% were not taking hypertension medications as prescribed. The full set of descriptive statistics appear in table 1.

    View this table:
    Table 1

    Descriptive characteristics of Philadelphia 9-digit ZIP codes (n=2002) and individual survey respondents (n=2083)

    Table 2 includes Spearman correlations between consumer credit and each covariate in the regression model. Higher area-level credit scores were positively moderately correlated with median age (r=0.38), percent with income >$100 000 (r=0.66), and percent with a graduate degree (r=0.56), and were negatively correlated with percent Black (r=−0.67). Count of pharmacies (r=0.01) and percent Hispanic/Latinx (r=0.03) were not correlated. Correlations between area-level credit scores and individual-level variables of age (r=0.01), Black race (r=−0.50), ethnicity (r=0.14) and health insurance (r=0.08) were null to moderate. Consumer credit scores were negatively correlated with an individual being in poverty (r=−0.38) and positively correlated with having an income >$100 000 (r=0.36).

    View this table:
    Table 2

    Spearman correlations between credit scores and regression model covariates

    Each 50-point increase in ZIP code credit score was associated with 8% lower risk of any chronic disease (PR=0.92; 95% CI 0.88 to 0.97); 6% lower risk of being overweight/obese (PR=0.94; 95% CI 0.90 to 0.99); 19% lower risk of diabetes (PR=0.81; 95% CI 0.69 to 0.94); 9% lower risk of having hypertension (PR=0.91; 95% CI 0.85 to 0.98) and 14% lower risk of smoking (PR=0.86; 95% CI: 0.75 to 0.99) (figure 1). Exercise, management of hypertension and pharmacy counts were not associated with credit scores. Compared with those in high prime credit areas, the total number of chronic disease, overweight/obesity, diabetes and hypertension were higher with each worsening credit tier, suggesting a dose–response relationship. Subprime credit was associated with a 15%–70% increased risk across the chronic diseases measured (figure 2). Full model results appear in the online supplementary appendix 1. There was no evidence of attenuation of race or racial composition when credit scores were included in any of the models.

    Supplementary file 1

    Figure 1
    Figure 1

    Adjusted PR for chronic disease outcomes and behaviours per 50-point FICO score increment.*All models adjusted for area-level and individual-level income, education, age, race, ethnicity, individual-level insurance status and area-level pharmacy counts. PR, prevalence ratio; HTN, hypertension.

    Figure 2
    Figure 2

    Adjusted PR for chronic disease outcomes and risk behaviours per increasing credit tier, compared to high prime (best credit). *All models adjusted for area-level and individual-level income, education, age, race, ethnicity, individual-level insurance status, area-level pharmacy counts and credit tier designations from the Federal Reserve Bank of New York.32 PR, prevalence ratio; HTN, hypertension.

    Conclusion

    This is the first study to examine area-level consumer credit as an exposure for chronic diseases and the health behaviours that shape those diseases. Findings suggest that area-level consumer credit may be incrementally associated with the presence of chronic diseases and some behavioural risk factors associated with disease, but not disease management behaviours. This may support the hypothesis that chronic disease is an economic shock that has sustained impact on economic well-being, which can have influences that expand beyond the individual. Alternately, our findings could indicate that areas with a higher prevalence of chronic disease may comprise many people who are facing greater economic challenges which may be reflected in area-level credit scores.

    There are several plausible explanations for how area-level credit may be related to individual health outcomes12:

    (1) Area-level credit scores could depict the context in which health and disease arise. Area-level consumer credit ratings may represent the aggregate effects of individual-level pathways (compositional factor) or an ecological feature of the social environment (contextual factor). Credit scores may reflect personal characteristics, such as an individual’s ability to manage complex processes, like navigating payment schedules, or preferences for risky behaviours. For example, studies have shown that credit scores are higher among those with greater speed and capacity for processing and responding to information,2 while another showed that individuals with higher bankruptcy risk were more likely to experience car crashes,35 and another estimated that people who were more able to delay gratification had FICO scores that were approximately 30 points higher than those least willing to delay gratification.36 While these are individual-level findings that cannot be imputed to area-level mechanisms, it is possible that area-level credit scores represent the compositional effects of people with certain characteristics clustering in certain areas. The present study’s findings of no association between credit and chronic disease management behaviours suggest that area-level consumer credit may not reflect that areas with poorer credit comprise people with a certain set of behavioural characteristics, as some have theorised for individual’s credit scores.2 3 36 Instead, as other recent work suggests,5 area-level consumer credit may better approximate a measure of area-level SEP.

    There may be stronger evidence to support our hypothesis that area-level credit scores represent socioeconomic contextual characteristics of an environment, as has been previously suggested.5 Additionally, area-level consumer credit ratings may alter the retail landscape through business loans and interest rates, and influence the placement of resources to manage the disease, such as retail pharmacies. However, the current study found no association between credit scores and presence of retail pharmacies, suggesting that area-level credit scores may not influence the placement of retailers that may be related to management of chronic disease behaviours. Future study should consider whether or not area-level credit scores are associated with the placement of other health-related resources.

    (2) Area-level consumer credit may also reflect areas where disease prevalence is high, especially when costs to manage or treat those chronic diseases may lead to populations who experience higher medical costs. For example, one ecological study in the USA found small but significant increases in city-wide influenza severity with a higher area-level credit card and mortgage default rates.4 The authors attributed this to the increased likelihood of consumer credit borrowing in areas with higher disease rates. Other recent work suggests that each SD increase in area-level credit score was associated with 26% greater odds of better individual self-rated health.5 The present study’s findings of higher risk of common chronic diseases further support this pathway.

    While this study suggests that area-level credit scores may be promising to consider in health studies, the inclusion of area-level consumer credit to understand health may pose challenges to health equity. Credit score models may be less accurate predictors of loan default for racial/ethnic minority individuals37 and communities,10 thus we adjusted for individual and area-level race and ethnicity in our models. Despite that the use of race and public assistance is prohibited by the Consumer Credit Protection Act, there is some evidence that low-value and subprime credit products are marketed to racial-ethnic minority communities,10 and that living in an area recovering from economic downturn can adversely affect credit.11 In Philadelphia, for example, the median credit score in predominantly White areas is 132 points higher than predominantly non-White areas.33 These communities are more likely to include micro-lenders, community-based and faith-based organisations, and credit unions, which would not contribute to credit scoring.10 Although we found no evidence of attenuation of the relationship between race, racial composition and health outcomes when credit was included in regression models, results may mask heterogeneity in the relationship between credit and health for specific subgroups. This should be explored in future studies.

    This cross-sectional study could not assess temporality and no information was available on length of credit history or the length of time with a chronic disease. It is possible that the same factors that give rise to credit may also be giving rise to health outcomes—unmeasured factors that are related to health might also be related to credit scores. In short, we cannot assume that one’s credit score is exogenous to one’s health outcomes and must consider that when interpreting results of studies linking credit and health.12 Measurement error is also a factor, given that 20% of Philadelphia residents have no record of a credit score with the Federal Reserve in 2015.32 Across the USA, approximately 20% of the population have no credit, either due to no records on account (11%; ‘credit invisible’) or insufficiently short or outdated histories (8%)38 39; however, the need for or usage of credit of itself may be associated with health. The measure cannot reflect behaviours or outcomes among those who do not yet have a credit score. Credit scores are not perfect predictors of future default behaviour and may be influenced by the credit products available in a particular area.10 Results may not be generalisable to other areas.

    This study contributes to the growing body of literature suggesting a relationship between consumer credit and health or disease and suggests a relationship between area-level credit and individual health. While the study highlights the need to understand causal mechanisms linking credit and disease, this analysis forms a foundation to explore mechanisms by which consumer credit might be related to chronic disease, opens a new path for considering which area-level factors might best be used to identify chronic disease patterns.

    What is already known on this subject

    • Consumer credit is increasingly linked to health outcomes, yet few studies have linked consumer credit scores and non-communicable or chronic disease.

    • No studies have assessed the behavioural mechanisms by which consumer credit is linked to chronic disease.

    What this study adds

    • This study identifies area-level consumer credit as a potential exposure for chronic disease outcomes, but less so for behaviours associated with increased chronic disease risk.

    • In doing so, it offers researchers a novel exposure to consider to better understand chronic disease patterns.

    References

    1. ↵
      1. Israel S,
      2. Caspi A,
      3. Belsky DW, et al
      . Credit scores, cardiovascular disease risk, and human capital. Proc Natl Acad Sci U S A 2014;111:17087–92. doi:10.1073/pnas.1409794111
      OpenUrlAbstract/FREE Full Text
    2. ↵
      1. Li Y,
      2. Gao J,
      3. Enkavi AZ, et al
      . Sound credit scores and financial decisions despite cognitive aging. Proc Natl Acad Sci U S A 2015;112:65–9. doi:10.1073/pnas.1413570112
      OpenUrlAbstract/FREE Full Text
    3. ↵
      1. Brockett PL,
      2. Golden LL
      . Biological and psychobehavioral correlates of credit scores and automobile insurance losses: toward an explication of why credit scoring works. Journal of Risk & Insurance 2007;74:23–63. doi:10.1111/j.1539-6975.2007.00201.x
      OpenUrl
    4. ↵
      1. Houle JN,
      2. Collins JM,
      3. Schmeiser MD
      . Flu and finances: influenza outbreaks and loan defaults in US Cities, 2004-2012. Am J Public Health 2015;105:e75–80. doi:10.2105/AJPH.2015.302671
      OpenUrl
    5. ↵
      1. Knapp EA,
      2. Dean LT
      . Consumer credit scores as a novel tool for identifying health in urban U.S. neighborhoods. Ann Epidemiol 2018;28:724–9. doi:10.1016/j.annepidem.2018.07.013
      OpenUrl
    6. ↵
      1. Dobkin C,
      2. Finkelstein A,
      3. Kluender R, et al
      . The economic consequences of hospital admissions. Am Econ Rev 2018;108:308–52. doi:10.1257/aer.20161038
      OpenUrl
    7. ↵
      1. Lee Y,
      2. Song I
      . A study on the economic factor associated with suicide: debt and suicide ideation. Mental Health & Social Work 2015;43:58–82.
      OpenUrl
    8. ↵
      Board of Governors of the Federal Reserve System. Report on the Economic Well-Being of U.S. Households in 2014. The Consumer and community development research section: Board of Governors of the Federal Reserve System, 2015:21–7.
    9. ↵
      1. Lyons AC,
      2. Rachlis M,
      3. Scherpf E
      . What’s in a score? Differences in consumers’ credit knowledge using OLS and Quantile Regressions. J Consum Aff 2007;41:223–49. doi:10.1111/j.1745-6606.2007.00079.x
      OpenUrlCrossRef
    10. ↵
      1. Rice L,
      2. Swesnik D
      . Discriminatory effects of credit scoring on communities of color. Suffolk UL Rev 2013;46:935.
      OpenUrl
    11. ↵
      1. AVERY R,
      2. CALEM P,
      3. CANNER G
      . Consumer credit scoring: Do situational circumstances matter? J Bank Financ 2004;28:835–56. doi:10.1016/S0378-4266(03)00202-4
      OpenUrl
    12. ↵
      1. Dean LT,
      2. Nicholas LH
      . Getting credit for your health? Using credit scores to understand predictors and consequences of disease. American Journal of Public Health. In Press.;2018.
    13. ↵
      1. Weiss R
      . Cost of a lymphedema treatment mandate-10 years of experience in the Commonwealth of Virginia. Health Econ Rev 2016;6:42. doi:10.1186/s13561-016-0117-3
      OpenUrl
    14. ↵
      1. Qato DM,
      2. Zenk S,
      3. Wilder J, et al
      . The availability of pharmacies in the United States: 2007-2015. PLoS One 2017;12:e0183172. doi:10.1371/journal.pone.0183172
    15. ↵
      1. Braveman P,
      2. Egerter S,
      3. Williams DR
      . The social determinants of health: coming of age. Annu Rev Public Health 2011;32:381–98. doi:10.1146/annurev-publhealth-031210-101218
      OpenUrlCrossRefPubMedWeb of Science
    16. ↵
      1. Krieger N
      . Epidemiology and social sciences: towards a critical reengagement in the 21st century. Epidemiol Rev 2000;22:155–63. doi:10.1093/oxfordjournals.epirev.a018014
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. Morris JN
      . Uses of epidemiology. Edinburgh: Churchill Livingstone 23 Ravelston Terrace, 1975.
    18. ↵
      1. Sydenstricker E
      . Health and environment: Ayer Company Pub, 1972.
    19. ↵
      1. Krieger N
      . A glossary for social epidemiology. J Epidemiol Community Health 2001;55:693–700. doi:10.1136/jech.55.10.693
      OpenUrlFREE Full Text
    20. ↵
      1. Galobardes B,
      2. Shaw M,
      3. Lawlor DA, et al
      . Indicators of socioeconomic position (part 1). J Epidemiol Community Health 2006;60:7–12. doi:10.1136/jech.2004.023531
      OpenUrlAbstract/FREE Full Text
    21. ↵
      1. Galobardes B,
      2. Shaw M,
      3. Lawlor DA, et al
      . Indicators of socioeconomic position (part 2). J Epidemiol Community Health 2006;60:95–101. doi:10.1136/jech.2004.028092
      OpenUrlAbstract/FREE Full Text
    22. ↵
      1. Krieger N,
      2. Williams DR,
      3. Moss NE
      . Measuring social class in US public health research: concepts, methodologies, and guidelines. Annu Rev Public Health 1997;18:341–78. doi:10.1146/annurev.publhealth.18.1.341
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. Zafar SY
      . Financial toxicity of cancer care: it’s time to intervene. J Natl Cancer Inst 2016;108:djv370. doi:10.1093/jnci/djv370
      OpenUrl
    24. ↵
      1. Hojman DA,
      2. Miranda Á,
      3. Ruiz-Tagle J
      . Debt trajectories and mental health. Soc Sci Med 2016;167:54–62. doi:10.1016/j.socscimed.2016.08.027
      OpenUrl
    25. ↵
      1. Banegas MP,
      2. Guy GP,
      3. de Moor JS, et al
      . For working-age cancer survivors, medical debt and bankruptcy create financial hardships. Health Aff 2016;35:54–61. doi:10.1377/hlthaff.2015.0830
      OpenUrlAbstract/FREE Full Text
    26. ↵
      1. Hamel L,
      2. Norton M,
      3. Pollitz K, et al
      . The burden of medical debt: results from the kaiser family foundation/New York Times Medical Bills Survey: Kaiser Family Foundation, 2016.
    27. ↵
      1. Turunen E,
      2. Hiilamo H
      . Health effects of indebtedness: a systematic review. BMC Public Health 2014;14:489. doi:10.1186/1471-2458-14-489
    28. ↵
      1. Zurlo KA,
      2. Yoon W,
      3. Kim H
      . Unsecured consumer debt and mental health outcomes in middle-aged and older Americans. J Gerontol B Psychol Sci Soc Sci 2014;69:461–9. doi:10.1093/geronb/gbu020
      OpenUrlCrossRefPubMed
    29. ↵
      1. Rukavina M
      . Medical debt and its relevance when assessing creditworthiness. Suffolk UL Rev 2013;46:967.
      OpenUrl
    30. ↵
      1. Traub A
      . Credit reports and employment: findings from the 2012 national survey on credit card debt of low-and Middle-Income Households. Suffolk UL Rev 2013;46:983.
      OpenUrl
    31. ↵
      1. Richardson T,
      2. Elliott P,
      3. Roberts R
      . The relationship between personal unsecured debt and mental and physical health: a systematic review and meta-analysis. Clin Psychol Rev 2013;33:1148–62. doi:10.1016/j.cpr.2013.08.009
      OpenUrlCrossRefPubMed
    32. ↵
      Federal Reserve Bank of New York/Equifax. Community credit: a new perspective on america’s communities: Federal Reserve Bank of New York/Equifax, 2017.
    33. ↵
      1. Ratcliffe C,
      2. Lou C,
      3. Elliott D, et al
      . Financial health of residents: a city-level dashboard. Philadelphia, PA: The Urban Institute, 2017.
    34. ↵
      Centers for Disease Control and Prevention. Faststats – statistics by topic, 2017.
    35. ↵
      1. Morrison E,
      2. Gupta A,
      3. Olson L, et al
      . Health and financial fragility: evidence from car crashes and consumer Bankruptcy, 2013.
    36. ↵
      1. Meier S,
      2. Sprenger CD
      . Time discounting predicts creditworthiness. Psychol Sci 2012;23:56–8. doi:10.1177/0956797611425931
      OpenUrlCrossRefPubMed
    37. ↵
      1. Mester LJ
      . What’s the point of credit scoring? Business review 1997;3:3–16.
      OpenUrl
    38. ↵
      1. Getter DE
      . Consumer and credit reporting, scoring, and related policy issues. Congressional Research Service 2015.
    39. ↵
      1. Brevoort KP GP,
      2. Kambara M
      . Data point: credit invisibles. CFPBOo R, ed. Consumer financial protection bureau, 2015.

    Footnotes

    • Contributors LTD and KV: contributed to the conception, writing and approval of the final manuscript. LTD: led the analysis, supported by DMQ and YR who provided data and guided analysis, with contributing analyses for specific health outcomes from EAK and SS. All authors contributed to the drafting and final approval of the manuscript.

    • Funding This work was supported by the National Cancer Institute grant K01CA184288; National Institute of Mental Health R25MH083620 (Lorraine T Dean); the Johns Hopkins University Center for AIDS Research grant P30AI094189 (Lorraine T Dean, Sevly Snguon); and the Sidney Kimmel Cancer Center grant P30CA006973 (Lorraine T Dean, Kala Visvanathan, Sevly Snguon). Emily Knapp was supported by the Clinical Research and Epidemiology in Diabetes and Endocrinology Training Grant (T32DK062707). Statistical consultation was provided by the National Center for Research Resources and the National Center for Advancing Translational Sciences (NCATS) of the National Institutes of Health through Grant Number 1UL1TR001079.

    • Competing interests None declared.

    • Patient consent Not required.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Correction notice This article has been corrected since it first published online. An error in the Abstract has been corrected.