Comparison of Childbirth Delivery Outcomes and Costs of Care Between Women Experiencing vs Not Experiencing Homelessness | Health Disparities | JAMA Network Open | JAMA Network
[Skip to Navigation]
Sign In
Figure 1.  Patient Outcomes
Patient Outcomes

Adjusted for patient-level variables (ie, age, race/ethnicity, payer, smoking history, multiple birth status, and Elixhauser comorbidities) and hospital fixed effects. The cesarean delivery model was additionally adjusted for previous cesarean delivery.

aP value adjusted for multiple comparisons using Benjamini-Hochberg false discovery rate was statistically significant at 5%. Adjusted P = .03.

Figure 2.  Delivery-Associated Hospitalization Costs
Delivery-Associated Hospitalization Costs

Adjusted for patient-level variables (ie, age, race/ethnicity, payer, smoking history, multiple birth status, and Elixhauser comorbidities). Costs were inflated to 2019 dollars using the gross domestic product price index.

aP value adjusted for multiple comparisons using Benjamini-Hochberg false discovery rate was statistically significant at 5%. Adjusted P = .02.

Table 1.  Patient Characteristics by Housing Status (Descriptive)
Patient Characteristics by Housing Status (Descriptive)
Table 2.  Hospital Characteristics by Housing Status (Descriptive)
Hospital Characteristics by Housing Status (Descriptive)
1.
Henry  M, Watt  R, Mahathey  A, Ouellette  J, Sitler  A; Abt Associates. The 2019 annual homeless assessment report (AHAR) to Congress. US Department of Housing and Urban Development. Accessed March 18, 2021. https://www.huduser.gov/portal/sites/default/files/pdf/2019-AHAR-Part-1.pdf
2.
National Coalition for the Homeless. How many people experience homelessness? Accessed March 18, 2021. http://nationalhomeless.org/wp-content/uploads/2017/09/HowManyPeople.pdf
3.
National Alliance to End Homelessness. Federal funding for homelessness programs. Accessed March 10,2020. https://endhomelessness.org/ending-homelessness/policy/federal-funding-homelessness-programs/
4.
City of Los Angeles. Confronting the crisis: homelessness. Accessed March 10, 2020. https://www.lamayor.org/Homelessness
5.
Bharel  M, Lin  WC, Zhang  J, O’Connell  E, Taube  R, Clark  RE.  Health care utilization patterns of homeless individuals in Boston: preparing for Medicaid expansion under the Affordable Care Act.   Am J Public Health. 2013;103(suppl 2):S311-S317. doi:10.2105/AJPH.2013.301421 PubMedGoogle ScholarCrossref
6.
US Institute of Medicine Committee on Health Care for Homeless People.  Homelessness, Health, and Human Needs: US National Academies Press; 1988.
7.
Baggett  TP, Hwang  SW, O’Connell  JJ,  et al.  Mortality among homeless adults in Boston: shifts in causes of death over a 15-year period.   JAMA Intern Med. 2013;173(3):189-195. doi:10.1001/jamainternmed.2013.1604 PubMedGoogle ScholarCrossref
8.
Kushel  MB, Vittinghoff  E, Haas  JS.  Factors associated with the health care utilization of homeless persons.   JAMA. 2001;285(2):200-206. doi:10.1001/jama.285.2.200 PubMedGoogle ScholarCrossref
9.
Kushel  MB, Perry  S, Bangsberg  D, Clark  R, Moss  AR.  Emergency department use among the homeless and marginally housed: results from a community-based study.   Am J Public Health. 2002;92(5):778-784. doi:10.2105/AJPH.92.5.778 PubMedGoogle ScholarCrossref
10.
Koh  KA, Racine  M, Gaeta  JM,  et al.  Health care spending and use among people experiencing unstable housing in the era of accountable care organizations.   Health Aff (Millwood). 2020;39(2):214-223. doi:10.1377/hlthaff.2019.00687 PubMedGoogle ScholarCrossref
11.
Welch-Lazoritz  ML, Whitbeck  LB, Armenta  BE.  Characteristics of mothers caring for children during episodes of homelessness.   Community Ment Health J. 2015;51(8):913-920. doi:10.1007/s10597-014-9794-8 PubMedGoogle ScholarCrossref
12.
Beal  AC, Redlener  I.  Enhancing perinatal outcome in homeless women: the challenge of providing comprehensive health care.   Semin Perinatol. 1995;19(4):307-313. doi:10.1016/S0146-0005(05)80046-1 PubMedGoogle ScholarCrossref
13.
Killion  CM.  Special health care needs of homeless pregnant women.   ANS Adv Nurs Sci. 1995;18(2):44-56. doi:10.1097/00012272-199512000-00006 PubMedGoogle ScholarCrossref
14.
Stein  JA, Andersen  R, Gelberg  L.  Applying the Gelberg-Andersen behavioral model for vulnerable populations to health services utilization in homeless women.   J Health Psychol. 2007;12(5):791-804. doi:10.1177/1359105307080612 PubMedGoogle ScholarCrossref
15.
Bloom  KC, Bednarzyk  MS, Devitt  DL, Renault  RA, Teaman  V, Van Loock  DM.  Barriers to prenatal care for homeless pregnant women.   J Obstet Gynecol Neonatal Nurs. 2004;33(4):428-435. doi:10.1177/0884217504266775 PubMedGoogle ScholarCrossref
16.
Richards  R, Merrill  RM, Baksh  L.  Health behaviors and infant health outcomes in homeless pregnant women in the United States.   Pediatrics. 2011;128(3):438-446. doi:10.1542/peds.2010-3491 PubMedGoogle Scholar
17.
Little  M, Shah  R, Vermeulen  MJ, Gorman  A, Dzendoletas  D, Ray  JG.  Adverse perinatal outcomes associated with homelessness and substance use in pregnancy.   CMAJ. 2005;173(6):615-618. doi:10.1503/cmaj.050406PubMedGoogle ScholarCrossref
18.
Stein  JA, Lu  MC, Gelberg  L.  Severity of homelessness and adverse birth outcomes.   Health Psychol. 2000;19(6):524-534. doi:10.1037/0278-6133.19.6.524PubMedGoogle ScholarCrossref
19.
Paterson  CM, Roderick  P.  Obstetric outcome in homeless women.   BMJ. 1990;301(6746):263-266. doi:10.1136/bmj.301.6746.263 PubMedGoogle ScholarCrossref
20.
Clark  RE, Weinreb  L, Flahive  JM, Seifert  RW.  Homelessness contributes to pregnancy complications.   Health Aff (Millwood). 2019;38(1):139-146. doi:10.1377/hlthaff.2018.05156 PubMedGoogle ScholarCrossref
21.
Healthcare Cost and Utilization Project (HCUP). HCUP databases. Agency for Healthcare Research and Quality. Accessed July 11, 2018. https://www.hcup-us.ahrq.gov/databases.jsp
22.
Karaca  Z, Wong  HS, Mutter  RL. Characteristics of Homeless and Non-Homeless Individuals Using Inpatient and Emergency Department Services, 2008: Statistical Brief #152. In:  Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. US Agency for Healthcare Research and Quality; 2013.
23.
Sun  R, Karaca  Z, Wong  HS. Characteristics of Homeless Individuals Using Emergency Department Services in 2014: Statistical Brief #229. In:  Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. US Agency for Healthcare Research and Quality; 2006.
24.
Manzano-Nunez  R, Zogg  CK, Bhulani  N,  et al.  Association of Medicaid expansion policy with outcomes in homeless patients requiring emergency general surgery.   World J Surg. 2019;43(6):1483-1489. doi:10.1007/s00268-019-04932-0 PubMedGoogle ScholarCrossref
25.
White  B, Ellis  C, Jones  W, Moran  W, Simpson  K.  The effect of the global financial crisis on preventable hospitalizations among the homeless in New York state.   J Health Serv Res Policy. 2018;23(2):80-86. doi:10.1177/1355819617742180 PubMedGoogle ScholarCrossref
26.
White  BM, Ellis  C  Jr, Simpson  KN.  Preventable hospital admissions among the homeless in California: a retrospective analysis of care for ambulatory care sensitive conditions.   BMC Health Serv Res. 2014;14:511. doi:10.1186/s12913-014-0511-7 PubMedGoogle ScholarCrossref
27.
AHA annual survey database. American Hospital Association. Accessed March 24, 2021. https://www.ahadata.com/aha-annual-survey-database
28.
Hospital 2552-2010 form. Centers for Medicare & Medicaid Services. Accessed March 24, 2021. https://www.cms.gov/Research-Statistics-Data-and-Systems/Downloadable-Public-Use-Files/Cost-Reports/Hospital-2010-form
29.
Kuklina  EV, Whiteman  MK, Hillis  SD,  et al.  An enhanced method for identifying obstetric deliveries: implications for estimating maternal morbidity.   Matern Child Health J. 2008;12(4):469-477. doi:10.1007/s10995-007-0256-6 PubMedGoogle ScholarCrossref
30.
Fingar  KR, Hambrick  MM, Heslin  KC,  et al. Trends and Disparities in Delivery Hospitalizations Involving Severe Maternal Morbidity, 2006-2015: Statistical Brief #243. In:  Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. US Agency for Healthcare Research and Quality; 2018. Accessed April 1, 2021. https://www.ncbi.nlm.nih.gov/books/NBK532465/
31.
Zhong  QY, Gelaye  B, Fricchione  GL, Avillach  P, Karlson  EW, Williams  MA.  Adverse obstetric and neonatal outcomes complicated by psychosis among pregnant women in the United States.   BMC Pregnancy Childbirth. 2018;18(1):120. doi:10.1186/s12884-018-1750-0 PubMedGoogle ScholarCrossref
32.
Hayward  RM, Foster  E, Tseng  ZH.  Maternal and fetal outcomes of admission for delivery in women with congenital heart disease.   JAMA Cardiol. 2017;2(6):664-671. doi:10.1001/jamacardio.2017.0283 PubMedGoogle ScholarCrossref
33.
Healthcare Cost and Utilization Project (HCUP). HCUP Elixhauser Comorbidity Software. Agency for Healthcare Research and Quality. Accessed July 11, 2018. https://www.hcup-us.ahrq.gov/toolssoftware/comorbidity/comorbidity.jsp
34.
Wiley  LK, Shah  A, Xu  H, Bush  WS.  ICD-9 tobacco use codes are effective identifiers of smoking status.   J Am Med Inform Assoc. 2013;20(4):652-658. doi:10.1136/amiajnl-2012-001557 PubMedGoogle ScholarCrossref
35.
Wooldridge  JM.  Econometric Analysis of Cross Section and Panel Data. MIT Press; 2001.
36.
Li  F, Morgan  KL, Zaslavsky  AM.  Balancing covariates via propensity score weighting.   Journal of the American Statistical Association. 2018;113(521):390-400. doi:10.1080/01621459.2016.1260466 PubMedGoogle ScholarCrossref
37.
Li  F, Thomas  LE, Li  F.  Addressing extreme propensity scores via the overlap weights.   Am J Epidemiol. 2019;188(1):250-257.PubMedGoogle Scholar
38.
Thomas  LE, Li  F, Pencina  MJ.  Overlap weighting: a propensity score method that mimics attributes of a randomized clinical trial.   JAMA. 2020;323(23):2417-2418. doi:10.1001/jama.2020.7819 PubMedGoogle ScholarCrossref
39.
Hellevik  O.  Linear versus logistic regression when the dependent variable is a dichotomy.   Qual Quant. 2009;43(1):59-74. doi:10.1007/s11135-007-9077-3 Google ScholarCrossref
40.
Williams  R.  Using the margins command to estimate and interpret adjusted predictions and marginal effects.   The Stata Journal. 2012;12(2):308-331. doi:10.1177/1536867X1201200209 Google ScholarCrossref
41.
Benjamini  Y, Hochberg  Y.  Controlling the false discovery rate: a practical and powerful approach to multiple testing.   J R Stat Soc Series B Stat Methodol. 1995; 57(1):289-300. doi:10.1111/j.2517-6161.1995.tb02031.x Google Scholar
42.
Eunice Kennedy Shriver National Institute of Child Health and Human Development. What causes preterm labor and birth? Updated January 31, 2017. Accessed February 22, 2021. https://www.nichd.nih.gov/health/topics/preterm/conditioninfo/causes
44.
Cleveland Clinic. Placental accreta. Accessed March 15, 2020. https://my.clevelandclinic.org/health/diseases/17846-placenta-accreta
45.
Tikkanen  M.  Placental abruption: epidemiology, risk factors and consequences.   Acta Obstet Gynecol Scand. 2011;90(2):140-149. doi:10.1111/j.1600-0412.2010.01030.x PubMedGoogle ScholarCrossref
46.
American College of Obstetricians and Gynecologists Committee on Practice Bulletins—Obstetrics.  ACOG practice bulletin No. 127: management of preterm labor.   Obstet Gynecol. 2012;119(6):1308-1317. doi:10.1097/AOG.0b013e31825af2f0 PubMedGoogle ScholarCrossref
47.
Barrett  PM, McCarthy  FP, Evans  M,  et al.  Stillbirth is associated with increased risk of long-term maternal renal disease: a nationwide cohort study.   Am J Obstet Gynecol. 2020;223(3):427.e1-427.e14. doi:10.1016/j.ajog.2020.02.031 PubMedGoogle ScholarCrossref
48.
Mitselou  N, Hallberg  J, Stephansson  O, Almqvist  C, Melén  E, Ludvigsson  JF.  Adverse pregnancy outcomes and risk of later allergic rhinitis-Nationwide Swedish cohort study.   Pediatr Allergy Immunol. 2020;31(5):471-479. doi:10.1111/pai.13230PubMedGoogle ScholarCrossref
49.
Centers for Medicare and Medicaid Services. Accountable Health Communities Model. Updated December 18, 2020. Accessed March 18, 2021. https://innovation.cms.gov/initiatives/ahcm/.
50.
Fraze  TK, Brewster  AL, Lewis  VA, Beidler  LB, Murray  GF, Colla  CH.  Prevalence of screening for food insecurity, housing instability, utility needs, transportation needs, and interpersonal violence by US physician practices and hospitals.   JAMA Netw Open. 2019;2(9):e1911514. doi:10.1001/jamanetworkopen.2019.11514 PubMedGoogle Scholar
51.
Kramer  MS.  The epidemiology of adverse pregnancy outcomes: an overview.   J Nutr. 2003;133(5)(suppl 2):1592S-1596S. doi:10.1093/jn/133.5.1592S PubMedGoogle ScholarCrossref
52.
Nyamathi  AM, Leake  B, Gelberg  L.  Sheltered versus nonsheltered homeless women differences in health, behavior, victimization, and utilization of care.   J Gen Intern Med. 2000;15(8):565-572. doi:10.1046/j.1525-1497.2000.07007.x PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Original Investigation
    Obstetrics and Gynecology
    April 22, 2021

    Comparison of Childbirth Delivery Outcomes and Costs of Care Between Women Experiencing vs Not Experiencing Homelessness

    Author Affiliations
    • 1Department of Health Policy and Management, University of California, Los Angeles Fielding School of Public Health
    • 2Department of Healthcare Economics, UnitedHealthcare, Cypress, California
    • 3Department of Family Medicine, David Geffen School of Medicine at University of California, Los Angeles
    • 4Office of Healthcare Transformation and Innovation, VA Greater Los Angeles Healthcare System, Los Angeles, California
    • 5University of California, Los Angeles Center for Health Policy Research
    • 6Division of General Internal Medicine and Health Services Research, David Geffen School of Medicine at University of California Los Angeles
    • 7Department of Public Health, Graduate School of Medicine, University of Tokyo, Bunkyo-ku, Tokyo, Japan
    JAMA Netw Open. 2021;4(4):e217491. doi:10.1001/jamanetworkopen.2021.7491
    Key Points

    Question  Do pregnant women experiencing homelessness have poorer childbirth delivery outcomes and higher costs of care compared with pregnant women not experiencing homelessness?

    Findings  In this multistate population-based cross-sectional study of 15 029 pregnant women experiencing homelessness and 308 242 pregnant women not experiencing homelessness, those experiencing homelessness were more likely to experience preterm labor and had higher delivery-associated costs. Those experiencing homelessness were also more likely to experience placental abnormalities, although this difference was not statistically significant.

    Meaning  These findings suggest that, within the same hospital, housing status is associated with preterm labor and higher delivery costs.

    Abstract

    Importance  Women and families constitute the fastest-growing segments of the homeless population. However, there is limited evidence on whether women experiencing homelessness have poorer childbirth delivery outcomes and higher costs of care compared with women not experiencing homelessness.

    Objective  To compare childbirth delivery outcomes and costs of care between pregnant women experiencing homelessness vs those not experiencing homelessness.

    Design, Setting, and Participants  This cross-sectional study included 15 029 pregnant women experiencing homelessness and 308 242 pregnant women not experiencing homelessness who had a delivery hospitalization in 2014. The study used statewide databases that included all hospital admissions in 3 states (ie, Florida, Massachusetts, and New York). Delivery outcomes and delivery-associated costs were compared between pregnant women experiencing homelessness and those not experiencing homelessness cared for at the same hospital (analyzed using the overlap propensity-score weighting method and multivariable regression models with hospital fixed effects). The Benjamini-Hochberg false discovery rate procedure was used to account for multiple comparisons. Data were analyzed from January 2020 through May 2020.

    Exposure  Housing status at delivery hospitalization.

    Main Outcomes and Measures  Outcome variables included obstetric complications (ie, antepartum hemorrhage, placental abnormalities, premature rupture of the membranes, preterm labor, and postpartum hemorrhage), neonatal complications (ie, fetal distress, fetal growth restriction, and stillbirth), delivery method (ie, cesarean delivery), and delivery-associated costs.

    Results  Among 15 029 pregnant women experiencing homelessness (mean [SD] age, 28.5 [5.9] years) compared with 308 242 pregnant women not experiencing homelessness (mean [SD] age, 29.4 [5.8] years) within the same hospital, those experiencing homelessness were more likely to experience preterm labor (adjusted probability, 10.5% vs 6.7%; adjusted risk difference [aRD], 3.8%; 95% CI, 1.2%-6.5%; adjusted P = .03) and had higher delivery-associated costs (adjusted costs, $6306 vs $5888; aRD, $417; 95% CI, $156-$680; adjusted P = .02) compared with women not experiencing homelessness. Those experiencing homelessness also had a higher probability of placental abnormalities (adjusted probability, 4.0% vs 2.0%; aRD, 1.9%; 95% CI, 0.4%-3.5%; adjusted P = .053), although this difference was not statistically significant.

    Conclusions and Relevance  This study found that women experiencing homelessness, compared with those not experiencing homelessness, who had a delivery and were admitted to the same hospital were more likely to experience preterm labor and incurred higher delivery-associated costs. These findings suggest wide disparities in delivery-associated outcomes between women experiencing homelessness and those not experiencing homelessness in the US. The findings highlight the importance for health care professionals to actively screen pregnant women for homelessness during prenatal care visits and coordinate their care with community health programs and social housing programs to make sure their health care needs are met.

    Introduction

    More than 560 000 individuals in the United States experience homeless on any given night,1 and each year, an estimated 1% of the US population experience homelessness.2 Homelessness increased in metropolitan cities in 2019 compared with 2018,1 and it has become a major concern that has received attention for funding for housing initiatives at the local and national levels.3,4 Individuals who are experiencing unstable housing or homelessness have higher disease burdens and mortality rates and 2.5-fold higher health care spending vs a comparable housed population.5-10

    While most people experiencing homelessness are men, approximately 1 in 4 individuals experiencing homelessness are women or girls.1 More importantly, women and families constitute the fastest-growing segment of the homeless population.11 Circumstances associated with homelessness, such as malnutrition, and preexisting conditions, including substance abuse and mental illness, are associated with poorer health outcomes of the mother and fetus.12,13 Women experiencing homelessness are more likely to have dire living situations, which pose additional challenges that limit the ability of these women to take care of their health or seek care.14

    There is limited evidence on whether pregnant women experiencing homelessness also experience poorer childbirth-associated outcomes and higher costs of care compared with women not experiencing homelessness. The current literature suggests that homelessness among pregnant women is associated with increased barriers to accessing prenatal care15 and fewer prenatal care visits.16 It is also associated with an increased likelihood of having infants who are born preterm, longer hospital lengths of stay, and increased likelihood of having babies who have low birth weight, are small for their gestational age, and need neonatal intensive care.16-18 However, existing research is limited to small studies conducted at a single hospital,17,19 city (ie, Los Angeles),18 or state (ie, Massachusetts).20 Therefore, whether their findings are generalizable to other regions of the US remains unclear. Moreover, prior studies compared women experiencing homelessness vs those not experiencing homelessness across hospitals; therefore, observed disparities in health outcomes may be associated with differences in the quality of hospitals from which women sought care.20 Thus, robust evidence on health outcomes and costs of care among these women using population-level data from multiple states is critically important. Given this context, using state-wide databases that include all hospital discharges from 3 large and diverse states, we examined the association between homelessness and obstetric, neonatal, and health service outcomes among all pregnant women who had a delivery hospitalization in 2014.

    Methods

    The University of California-Los Angeles institutional review board deemed this cross-sectional study to not be human subject research and therefore exempt from institutional review board approval and informed consent. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

    Data Sources and Study Sample

    We analyzed the 2014 State Inpatient Database (SID) and the State Emergency Department Databases for 3 states (ie, Florida, Massachusetts, and New York) made available by the Healthcare Cost and Utilization Project of the Agency for Healthcare Research and Quality21 (see eAppendix in the Supplement for further discussion of these databases and excluded states). For each inpatient and hospital-affiliated emergency department discharge, there is an indicator for each patient's housing status reported directly from the hospitals, and this variable has been used in previous studies to identify the housing status of patients.22-26 Additionally, we used the 2016 American Hospital Association's Annual Survey Database27 and the 2014 Medicare Cost Reports28 for data on hospital characteristics.

    The study population was restricted to women ages 18 to 44 years who had a hospitalization for delivery in 2014. The analysis was restricted to the first hospitalization. We excluded women who were missing housing status or any of the key adjustment variables described in a following section. Childbirth deliveries were identified using the enhanced method for identifying deliveries as described elsewhere.29,30 This method uses a combination of International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9) diagnosis and procedure codes and diagnosis-related group codes and has been found to be less likely to miss severe obstetric complications compared with a standard method of using V27 codes.29 See eTable 1 in the Supplement for a list of codes.

    Outcome Variables

    Outcome variables were classified into 4 groups: (1) obstetric complications during pregnancy (ie, antepartum hemorrhage and placental abnormalities) or associated with labor (ie, premature rupture of membranes, preterm labor, and postpartum hemorrhage), (2) neonatal complications (ie, fetal distress, fetal growth restriction, and stillbirth), (3) delivery method (ie, cesarean delivery), and (4) delivery-associated costs. Delivery-associated costs were defined as the total cost of care of the delivery hospitalization. We used Healthcare Cost and Utilization Project’s cost-to-charge ratio file to calculate costs from total charges available in the SID. All costs were inflated to 2019 dollars using the gross domestic product price index. Childbirth delivery outcomes have been defined in prior studies.31,32 See eTable 2 in the Supplement for a list of codes.

    Adjustment Variables

    In our multivariable regression models, we adjusted for patient-level adjustment variables and hospital fixed effects. Patient-level adjustment variables included age (categorized as 18-24 years, 25-29 years, 30-34 years, 35-39 years, and 40-44 years), race/ethnicity (ie, non-Hispanic White, non-Hispanic Black, Hispanic, and other race/ethnicity), primary insurance type (ie, Medicare, Medicaid, private, self-pay/no charge/other), smoking history (ie, ever smoked), multiple birth status, and 26 comorbidities included in the Elixhauser Comorbidities Index33 (similar medical conditions were combined while ulcer and AIDS were excluded from the index owing to low cell counts [ie, <3]). Smoking status was identified using ICD-9 codes because these codes have been found to be reliable indicators associated with smoking status.34 In the cesarean delivery outcome model, we additionally adjusted for previous cesarean delivery. Hospital fixed effects (ie, indicator variables for each hospital) account for both measured and unmeasured (ie, time-invariant) characteristics of hospitals, allowing us to compare women experiencing and not experiencing homelessness within the same hospital.35

    Statistical Analysis

    To control for sample selection bias, we used a doubly robust overlap propensity score–weighing method.36,37 In the propensity score and the outcome regression models, we included all patient-level adjustment variables explained in the previous section. The outcome regression models additionally adjusted for hospital fixed effects.

    We used the overlap propensity score–weighting method, a new form of balancing weights using propensity scores, given that the distributions of propensity score showed limited overlap and extreme propensity scores. That is, there were small propensity scores for some individuals experiencing homelessness, which made it difficult to find the matched control in propensity score–matching method and led to inflated inverse weights when using inverse probability of treatment weights.36-38 The overlap weighting method upweights observations with the largest overlap in observed characteristics between 2 groups while downweighting those observations with the least overlap. In other words, it deemphasizes observations with propensity scores close to 0 or 1 and emphasizes those close to a propensity score of 0.5. An advantage of using this method over trimming methods is that it does not rely on arbitrarily selecting a cutoff point for the weights and does not involve removing observations. These weights provide estimates of the average treatment effect in the overlap population.37

    Next, we used multivariable linear probability models with Huber-White heteroscedasticity-robust standard errors. We could not use logistic regression models because small cell sizes for some combinations of patient characteristics resulted in complete or quasicomplete separation.39 Delivery-associated costs were analyzed using generalized linear models with log-link and gamma distribution. We adjusted for patient-level variables (ie, age, race/ethnicity, primary payer, smoking history, multiple birth status, and Elixhauser comorbidities) and hospital fixed effects. After fitting the regression models, adjusted outcomes were calculated using the marginal standardization form of predictive margins (also known as predictive margins or margins of responses); for each individual, we calculated estimated probabilities of each outcome with housing status fixed at each category (ie, 0 or 1) and then found the mean over the distribution of covariates in our sample.40

    We conducted several sensitivity analyses. First, because women experiencing homelessness are more likely to have limited and fragmented access to health care and thus may be less likely to have coded diagnoses of comorbidities, we conducted analyses accounting for these differences. To investigate how differential coding patterns between women experiencing and not experiencing homelessness may have been associated with changes to our findings, we reanalyzed the data without adjusting for Elixhauser comorbidities in our regression models. Second, given that women not experiencing homelessness may differ from those experiencing homelessness in many ways that cannot be accounted for by adjusting for observable characteristics, we used a comparison group consisting of women not experiencing homelessness who were in the lowest quartile of median household income by zip code (ie, who were low-income housed). Third, we examined whether outcomes differed between women experiencing and not experiencing homelessness when they were compared across hospitals using multivariable regression models without hospital fixed effects. Fourth, to determine how different modeling approaches may be associated with the magnitude and direction of delivery-associated costs, these costs were separately analyzed using negative binomial regression (ie, a log-link, negative binomial generalized linear model) and ordinary least squares with Huber-White robust heteroscedasticity standard errors.

    All analyses were conducted in SAS Enterprise Guide statistical software version 4.2 (SAS Institute) and Stata statistical software version 14 (StataCorp). Because our study had multiple outcome variables, we accounted for multiple comparisons using the Benjamini-Hochberg procedure, setting the false discovery rate threshold to q < .05.41 This method controls the expected proportion of false positives to less than 5% and is less conservative compared with family-wise error rate controlling procedures, such as the Bonferroni procedure. Reported P values are adjusted and 2-sided. Data were analyzed from January 2020 through May 2020.

    Results

    Our final sample consisted of 15 029 pregnant women experiencing homelessness and 308 242 pregnant women not experiencing homelessness. Women experiencing homelessness, compared with women not experiencing homelessness, were younger (mean [SD] age, 28.5 [5.9] years vs 29.4 [5.8] years); more likely to have Medicaid as the primary payer (14 695 women [97.8%] vs 145 264 women [47.1%]); and more likely to be non-Hispanic Black (4906 women [32.6%] vs 57 560 women [18.7%]), Hispanic (5495 women [36.6%] vs 57 315 women [18.6%]), or other race/ethnicity (3790 women [25.2%] vs 47 432 women [15.4%]). Women experiencing homelessness were also more likely to have coded diagnoses of hypertension, coagulopathy, obesity, alcohol abuse or liver disease, and psychoses. They were less likely to have ever been a smoker, have multiple births, and have coded diagnoses of hypothyroidism, deficiency anemias, or depression (Table 1). Women experiencing homelessness were more likely to be seen in small hospitals (ie, 1-99 beds) to medium-size hospitals (ie, 100-300 beds), hospitals with minor teaching status, government-owned hospitals, and hospitals most likely to be safety net hospitals (Table 2).

    Association Between Homelessness and Delivery Outcomes

    After adjusting for patient characteristics and hospital fixed effects, we found that women experiencing homelessness, vs those not experiencing homelessness, had a higher likelihood of preterm labor (adjusted probability, 10.5% vs 6.7%; adjusted risk difference [aRD], 3.8%; 95% CI, 1.2%-6.5%; adjusted P = .03) and had higher delivery-associated costs (adjusted costs, $6306 vs $5888; aRD, $417; 95% CI; $156-$680; adjusted P = .02). Women experiencing homelessness also had a higher probability of placental abnormalities, although this difference was not statistically significant (adjusted probability, 4.0% vs 2.0%; aRD, 1.9%; 95% CI, 0.4%-3.5%; adjusted P = .053) (Figure 1 and Figure 2; eTable 3 in the Supplement).

    Sensitivity Analyses

    Our findings were qualitatively unchanged by the analysis, excluding Elixhauser comorbidities, or by the analysis comparing women experiencing homelessness with women not experiencing homelessness who were low income (eTable 4 and eTable 5 in the Supplement). When we compared women experiencing homelessness with women not experiencing homelessness across hospitals (using weighted regression models without hospital fixed effects), we found that those experiencing homelessness had a lower likelihood of a cesarean delivery but a higher likelihood of fetal distress. We found no evidence that delivery-associated costs differed between the 2 groups comparing across hospitals (Figure 2; eTable 6 in the Supplement). We obtained a similar delivery-associated cost outcome with a negative binomial model and a greater cost difference with an ordinary least squares model (eTable 7 in the Supplement).

    Discussion

    Using statewide data sets of all hospital admissions from 3 large and diverse states, this cross-sectional study found that pregnant women experiencing homelessness had a higher likelihood of experiencing preterm labor and higher delivery-associated costs compared with pregnant women not experiencing homelessness, even when they were cared for within the same hospital. Pregnant women experiencing homelessness also had a higher probability of placental abnormalities, although this difference was not statistically significant.

    Given that we compared women experiencing homelessness with those not experiencing homelessness at the same hospital, observed differences in delivery outcomes and costs may be associated with disparities in the quality of care these 2 groups of women received at the same hospital, as well as unmeasured risk factors of women experiencing homelessness. For instance, the higher preterm labor rates we observed among women experiencing homelessness may be associated with other unmeasured clinical characteristics, such as infections, stressful life events, drug and alcohol use, history of preterm labor, and shorter time intervals between pregnancies, among others.42 Known risk factors for placental abnormalities include cocaine use, having a placental previa from a previous pregnancy, and previous uterine surgery, among others.43 Given that women experiencing homelessness are less likely to receive prenatal care16 and more likely to abuse alcohol and illicit drugs,12,13 it is possible that these women are presenting with heightened risk at the time of delivery.

    Placental abnormalities (ie, previa, abruptio, and accreta) can lead to further complications, such as heavy bleeding, preterm birth, and stillbirth,44,45 and preterm labor can result in preterm births.46 Our findings were consistent with single-facility studies that found that mothers experiencing homelessness were more likely than those not experiencing homelessness to have adverse delivery outcomes, such as preterm delivery, an infant with low birth weight, and an infant who was small for gestational age.17,19 Such adverse delivery outcomes can be associated with detrimental outcomes for the mother and the infant over the long term.47,48

    Our findings that women experiencing homelessness had worse delivery-associated outcomes suggest that these women need to be identified and be connected to resources to help springboard them out of homelessness. The Centers for Medicaid & Medicare Services established the Accountable Health Communities Model, which provides funding for selected clinic-community collaborations to address social determinants of health through efforts such as screening patients at clinics for social needs.49 However, less than a quarter of physicians screen patients for all 5 social needs: food insecurity, housing instability, unmet needs for utility, transportation needs, and interpersonal violence.50 Health care policies should encourage clinicians to screen pregnant women for social needs and collaborate with community resources, such as local housing authorities, so that pregnant women’s housing needs are met and they are able to receive care for their unmet health care needs.

    There are several plausible explanations for why we did not observe significant differences for some of the outcomes we studied (ie, stillbirth, antepartum hemorrhage, fetal growth restriction, and postpartum hemorrhage). First, they are rarer events compared with preterm labor, and therefore, our study may be underpowered to detect significant differences in the rate of these outcomes. It is also possible that some outcomes are associated largely with the quality of care of the hospital where the women delivered (with small within-hospital variations), with no differences to be observed when we compared women experiencing homelessness vs not experiencing homelessness at the same hospital. This hypothesis is supported by our finding from the sensitivity analysis excluding hospital fixed effects of higher rates of fetal distress and lower rates of cesarean delivery for women experiencing homelessness. Furthermore, women experiencing homelessness had higher delivery-associated costs compared with women not experiencing homelessness within the same hospitals but not compared across hospitals. These findings suggest that women experiencing homelessness may have received delivery-associated care in hospitals that have lower mean delivery costs, higher fetal distress rates, and lower cesarean delivery rates, perhaps associated with these institutions’ limited resources and the high-risk populations they serve.

    To our knowledge, this is one of few studies using multiple states to compare childbirth delivery outcomes and costs of care between women experiencing homelessness and those not experiencing homelessness. Prior studies16-18 have found that pregnant women experiencing homelessness were more likely to have adverse delivery outcomes, such as giving birth to infants who are born preterm, have low birth weight, or are small for gestational age. However, most existing studies17-20 have been limited in sample size and were conducted in a single state, city, or hospital. A population-level study16 using the Pregnancy Risk Assessment Monitoring System survey database from 31 states found that infants born to women experiencing homelessness had worse neonatal outcomes compared with infants born to women not experiencing homelessness. However, the study did not assess obstetric delivery complications or link survey data to hospital admission data. In our study, we included all hospital admission data from 3 states and compared obstetric, fetal, and health service outcomes. Our statistical models adjusted for hospital fixed effects, which allowed us to make robust within-hospital comparisons. Additionally, we used a doubly robust overlap propensity score–weighting method, which has the advantage of selecting a clinically relevant target population, achieving covariate balance between groups, and improving precision compared with other propensity score methods.38

    Limitations

    This study has several limitations. First, as for any cross-sectional study, the temporality between exposure and outcome could not be assessed. However, women were identified as experiencing homelessness at the time of discharge in our data; therefore, it is unlikely that their delivery complications led them to become homeless. Second, with administrative data, exposure and outcome misclassifications are possible. For instance, the severity of homelessness may not be captured fully in our data. However, if women experiencing homelessness for a short time were coded as experiencing homelessness long term or if housing status was underreported, this would bias our estimates toward the null. In addition, concerns about the overconcentration of individuals experiencing homelessness in some hospitals were addressed by the inclusion of hospital fixed effects. Third, while we adjusted for a broad set of patient demographic and clinical variables in our analyses, we were unable to adjust for other potential confounders, such as maternal body mass index, history of preterm birth, and number of prenatal care visits.51 Fourth, while our study includes all women experiencing homelessness and not experiencing homelessness who had a delivery hospitalization from 3 states, our findings may not be generalizable to individuals experiencing homelessness in states not included in our analysis or individuals who were not hospitalized for deliveries. We were limited by the availability of the state data and homeless indicator data in states that were not included in our study. On the West Coast, individuals experiencing homelessness are able to live outside of shelters year round owing to milder climates, and therefore may not be able to access social services as readily as these individuals living in other states.52

    Conclusions

    Using state-wide databases from 3 large states, we found that pregnant women experiencing homelessness were more likely to have preterm labor and had higher costs of delivery compared with pregnant women not experiencing homelessness who were cared for at the same hospital. These findings emphasize the importance of hospitals and clinicians to screen pregnant patients for social needs, including housing, and leverage community resources as soon as possible so that mothers can prioritize their health for safe delivery.

    Back to top
    Article Information

    Accepted for Publication: March 4, 2021.

    Published: April 22, 2021. doi:10.1001/jamanetworkopen.2021.7491

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Yamamoto A et al. JAMA Network Open.

    Corresponding Author: Ayae Yamamoto, PhD, Department of Healthcare Economics, UnitedHealthcare, 5701 Katella Ave, Cypress, CA 90630 (ayaey5@gmail.com).

    Author Contributions: Dr Yamamoto had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Yamamoto, Gelberg, Needleman, Miyawaki, Tsugawa.

    Acquisition, analysis, or interpretation of data: Yamamoto, Needleman, Kominski, Vangala, Miyawaki, Tsugawa.

    Drafting of the manuscript: Yamamoto, Tsugawa.

    Critical revision of the manuscript for important intellectual content: All authors.

    Statistical analysis: Yamamoto, Vangala, Miyawaki.

    Obtained funding: Yamamoto.

    Administrative, technical, or material support: Yamamoto, Tsugawa.

    Supervision: Gelberg, Needleman, Kominski, Tsugawa.

    Conflict of Interest Disclosures: Dr Tsugawa was supported by grants from the National Institutes of Health (NIH) National Institute on Minority Health and Health Disparities and NIH National Institute on Aging for other work not related to this study. No other disclosures were reported.

    Funding/Support: This research was supported by St Luke's International University. Dr Yamamoto was supported by grant No. TL1TR001883 from the National Institutes of Health National Center for Advancing Translational Science through the University of California, Los Angeles (UCLA) Clinical and Translational Science Institute and the UCLA Department of Health Policy and Management.

    Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the funders.

    References
    1.
    Henry  M, Watt  R, Mahathey  A, Ouellette  J, Sitler  A; Abt Associates. The 2019 annual homeless assessment report (AHAR) to Congress. US Department of Housing and Urban Development. Accessed March 18, 2021. https://www.huduser.gov/portal/sites/default/files/pdf/2019-AHAR-Part-1.pdf
    2.
    National Coalition for the Homeless. How many people experience homelessness? Accessed March 18, 2021. http://nationalhomeless.org/wp-content/uploads/2017/09/HowManyPeople.pdf
    3.
    National Alliance to End Homelessness. Federal funding for homelessness programs. Accessed March 10,2020. https://endhomelessness.org/ending-homelessness/policy/federal-funding-homelessness-programs/
    4.
    City of Los Angeles. Confronting the crisis: homelessness. Accessed March 10, 2020. https://www.lamayor.org/Homelessness
    5.
    Bharel  M, Lin  WC, Zhang  J, O’Connell  E, Taube  R, Clark  RE.  Health care utilization patterns of homeless individuals in Boston: preparing for Medicaid expansion under the Affordable Care Act.   Am J Public Health. 2013;103(suppl 2):S311-S317. doi:10.2105/AJPH.2013.301421 PubMedGoogle ScholarCrossref
    6.
    US Institute of Medicine Committee on Health Care for Homeless People.  Homelessness, Health, and Human Needs: US National Academies Press; 1988.
    7.
    Baggett  TP, Hwang  SW, O’Connell  JJ,  et al.  Mortality among homeless adults in Boston: shifts in causes of death over a 15-year period.   JAMA Intern Med. 2013;173(3):189-195. doi:10.1001/jamainternmed.2013.1604 PubMedGoogle ScholarCrossref
    8.
    Kushel  MB, Vittinghoff  E, Haas  JS.  Factors associated with the health care utilization of homeless persons.   JAMA. 2001;285(2):200-206. doi:10.1001/jama.285.2.200 PubMedGoogle ScholarCrossref
    9.
    Kushel  MB, Perry  S, Bangsberg  D, Clark  R, Moss  AR.  Emergency department use among the homeless and marginally housed: results from a community-based study.   Am J Public Health. 2002;92(5):778-784. doi:10.2105/AJPH.92.5.778 PubMedGoogle ScholarCrossref
    10.
    Koh  KA, Racine  M, Gaeta  JM,  et al.  Health care spending and use among people experiencing unstable housing in the era of accountable care organizations.   Health Aff (Millwood). 2020;39(2):214-223. doi:10.1377/hlthaff.2019.00687 PubMedGoogle ScholarCrossref
    11.
    Welch-Lazoritz  ML, Whitbeck  LB, Armenta  BE.  Characteristics of mothers caring for children during episodes of homelessness.   Community Ment Health J. 2015;51(8):913-920. doi:10.1007/s10597-014-9794-8 PubMedGoogle ScholarCrossref
    12.
    Beal  AC, Redlener  I.  Enhancing perinatal outcome in homeless women: the challenge of providing comprehensive health care.   Semin Perinatol. 1995;19(4):307-313. doi:10.1016/S0146-0005(05)80046-1 PubMedGoogle ScholarCrossref
    13.
    Killion  CM.  Special health care needs of homeless pregnant women.   ANS Adv Nurs Sci. 1995;18(2):44-56. doi:10.1097/00012272-199512000-00006 PubMedGoogle ScholarCrossref
    14.
    Stein  JA, Andersen  R, Gelberg  L.  Applying the Gelberg-Andersen behavioral model for vulnerable populations to health services utilization in homeless women.   J Health Psychol. 2007;12(5):791-804. doi:10.1177/1359105307080612 PubMedGoogle ScholarCrossref
    15.
    Bloom  KC, Bednarzyk  MS, Devitt  DL, Renault  RA, Teaman  V, Van Loock  DM.  Barriers to prenatal care for homeless pregnant women.   J Obstet Gynecol Neonatal Nurs. 2004;33(4):428-435. doi:10.1177/0884217504266775 PubMedGoogle ScholarCrossref
    16.
    Richards  R, Merrill  RM, Baksh  L.  Health behaviors and infant health outcomes in homeless pregnant women in the United States.   Pediatrics. 2011;128(3):438-446. doi:10.1542/peds.2010-3491 PubMedGoogle Scholar
    17.
    Little  M, Shah  R, Vermeulen  MJ, Gorman  A, Dzendoletas  D, Ray  JG.  Adverse perinatal outcomes associated with homelessness and substance use in pregnancy.   CMAJ. 2005;173(6):615-618. doi:10.1503/cmaj.050406PubMedGoogle ScholarCrossref
    18.
    Stein  JA, Lu  MC, Gelberg  L.  Severity of homelessness and adverse birth outcomes.   Health Psychol. 2000;19(6):524-534. doi:10.1037/0278-6133.19.6.524PubMedGoogle ScholarCrossref
    19.
    Paterson  CM, Roderick  P.  Obstetric outcome in homeless women.   BMJ. 1990;301(6746):263-266. doi:10.1136/bmj.301.6746.263 PubMedGoogle ScholarCrossref
    20.
    Clark  RE, Weinreb  L, Flahive  JM, Seifert  RW.  Homelessness contributes to pregnancy complications.   Health Aff (Millwood). 2019;38(1):139-146. doi:10.1377/hlthaff.2018.05156 PubMedGoogle ScholarCrossref
    21.
    Healthcare Cost and Utilization Project (HCUP). HCUP databases. Agency for Healthcare Research and Quality. Accessed July 11, 2018. https://www.hcup-us.ahrq.gov/databases.jsp
    22.
    Karaca  Z, Wong  HS, Mutter  RL. Characteristics of Homeless and Non-Homeless Individuals Using Inpatient and Emergency Department Services, 2008: Statistical Brief #152. In:  Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. US Agency for Healthcare Research and Quality; 2013.
    23.
    Sun  R, Karaca  Z, Wong  HS. Characteristics of Homeless Individuals Using Emergency Department Services in 2014: Statistical Brief #229. In:  Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. US Agency for Healthcare Research and Quality; 2006.
    24.
    Manzano-Nunez  R, Zogg  CK, Bhulani  N,  et al.  Association of Medicaid expansion policy with outcomes in homeless patients requiring emergency general surgery.   World J Surg. 2019;43(6):1483-1489. doi:10.1007/s00268-019-04932-0 PubMedGoogle ScholarCrossref
    25.
    White  B, Ellis  C, Jones  W, Moran  W, Simpson  K.  The effect of the global financial crisis on preventable hospitalizations among the homeless in New York state.   J Health Serv Res Policy. 2018;23(2):80-86. doi:10.1177/1355819617742180 PubMedGoogle ScholarCrossref
    26.
    White  BM, Ellis  C  Jr, Simpson  KN.  Preventable hospital admissions among the homeless in California: a retrospective analysis of care for ambulatory care sensitive conditions.   BMC Health Serv Res. 2014;14:511. doi:10.1186/s12913-014-0511-7 PubMedGoogle ScholarCrossref
    27.
    AHA annual survey database. American Hospital Association. Accessed March 24, 2021. https://www.ahadata.com/aha-annual-survey-database
    28.
    Hospital 2552-2010 form. Centers for Medicare & Medicaid Services. Accessed March 24, 2021. https://www.cms.gov/Research-Statistics-Data-and-Systems/Downloadable-Public-Use-Files/Cost-Reports/Hospital-2010-form
    29.
    Kuklina  EV, Whiteman  MK, Hillis  SD,  et al.  An enhanced method for identifying obstetric deliveries: implications for estimating maternal morbidity.   Matern Child Health J. 2008;12(4):469-477. doi:10.1007/s10995-007-0256-6 PubMedGoogle ScholarCrossref
    30.
    Fingar  KR, Hambrick  MM, Heslin  KC,  et al. Trends and Disparities in Delivery Hospitalizations Involving Severe Maternal Morbidity, 2006-2015: Statistical Brief #243. In:  Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. US Agency for Healthcare Research and Quality; 2018. Accessed April 1, 2021. https://www.ncbi.nlm.nih.gov/books/NBK532465/
    31.
    Zhong  QY, Gelaye  B, Fricchione  GL, Avillach  P, Karlson  EW, Williams  MA.  Adverse obstetric and neonatal outcomes complicated by psychosis among pregnant women in the United States.   BMC Pregnancy Childbirth. 2018;18(1):120. doi:10.1186/s12884-018-1750-0 PubMedGoogle ScholarCrossref
    32.
    Hayward  RM, Foster  E, Tseng  ZH.  Maternal and fetal outcomes of admission for delivery in women with congenital heart disease.   JAMA Cardiol. 2017;2(6):664-671. doi:10.1001/jamacardio.2017.0283 PubMedGoogle ScholarCrossref
    33.
    Healthcare Cost and Utilization Project (HCUP). HCUP Elixhauser Comorbidity Software. Agency for Healthcare Research and Quality. Accessed July 11, 2018. https://www.hcup-us.ahrq.gov/toolssoftware/comorbidity/comorbidity.jsp
    34.
    Wiley  LK, Shah  A, Xu  H, Bush  WS.  ICD-9 tobacco use codes are effective identifiers of smoking status.   J Am Med Inform Assoc. 2013;20(4):652-658. doi:10.1136/amiajnl-2012-001557 PubMedGoogle ScholarCrossref
    35.
    Wooldridge  JM.  Econometric Analysis of Cross Section and Panel Data. MIT Press; 2001.
    36.
    Li  F, Morgan  KL, Zaslavsky  AM.  Balancing covariates via propensity score weighting.   Journal of the American Statistical Association. 2018;113(521):390-400. doi:10.1080/01621459.2016.1260466 PubMedGoogle ScholarCrossref
    37.
    Li  F, Thomas  LE, Li  F.  Addressing extreme propensity scores via the overlap weights.   Am J Epidemiol. 2019;188(1):250-257.PubMedGoogle Scholar
    38.
    Thomas  LE, Li  F, Pencina  MJ.  Overlap weighting: a propensity score method that mimics attributes of a randomized clinical trial.   JAMA. 2020;323(23):2417-2418. doi:10.1001/jama.2020.7819 PubMedGoogle ScholarCrossref
    39.
    Hellevik  O.  Linear versus logistic regression when the dependent variable is a dichotomy.   Qual Quant. 2009;43(1):59-74. doi:10.1007/s11135-007-9077-3 Google ScholarCrossref
    40.
    Williams  R.  Using the margins command to estimate and interpret adjusted predictions and marginal effects.   The Stata Journal. 2012;12(2):308-331. doi:10.1177/1536867X1201200209 Google ScholarCrossref
    41.
    Benjamini  Y, Hochberg  Y.  Controlling the false discovery rate: a practical and powerful approach to multiple testing.   J R Stat Soc Series B Stat Methodol. 1995; 57(1):289-300. doi:10.1111/j.2517-6161.1995.tb02031.x Google Scholar
    42.
    Eunice Kennedy Shriver National Institute of Child Health and Human Development. What causes preterm labor and birth? Updated January 31, 2017. Accessed February 22, 2021. https://www.nichd.nih.gov/health/topics/preterm/conditioninfo/causes
    44.
    Cleveland Clinic. Placental accreta. Accessed March 15, 2020. https://my.clevelandclinic.org/health/diseases/17846-placenta-accreta
    45.
    Tikkanen  M.  Placental abruption: epidemiology, risk factors and consequences.   Acta Obstet Gynecol Scand. 2011;90(2):140-149. doi:10.1111/j.1600-0412.2010.01030.x PubMedGoogle ScholarCrossref
    46.
    American College of Obstetricians and Gynecologists Committee on Practice Bulletins—Obstetrics.  ACOG practice bulletin No. 127: management of preterm labor.   Obstet Gynecol. 2012;119(6):1308-1317. doi:10.1097/AOG.0b013e31825af2f0 PubMedGoogle ScholarCrossref
    47.
    Barrett  PM, McCarthy  FP, Evans  M,  et al.  Stillbirth is associated with increased risk of long-term maternal renal disease: a nationwide cohort study.   Am J Obstet Gynecol. 2020;223(3):427.e1-427.e14. doi:10.1016/j.ajog.2020.02.031 PubMedGoogle ScholarCrossref
    48.
    Mitselou  N, Hallberg  J, Stephansson  O, Almqvist  C, Melén  E, Ludvigsson  JF.  Adverse pregnancy outcomes and risk of later allergic rhinitis-Nationwide Swedish cohort study.   Pediatr Allergy Immunol. 2020;31(5):471-479. doi:10.1111/pai.13230PubMedGoogle ScholarCrossref
    49.
    Centers for Medicare and Medicaid Services. Accountable Health Communities Model. Updated December 18, 2020. Accessed March 18, 2021. https://innovation.cms.gov/initiatives/ahcm/.
    50.
    Fraze  TK, Brewster  AL, Lewis  VA, Beidler  LB, Murray  GF, Colla  CH.  Prevalence of screening for food insecurity, housing instability, utility needs, transportation needs, and interpersonal violence by US physician practices and hospitals.   JAMA Netw Open. 2019;2(9):e1911514. doi:10.1001/jamanetworkopen.2019.11514 PubMedGoogle Scholar
    51.
    Kramer  MS.  The epidemiology of adverse pregnancy outcomes: an overview.   J Nutr. 2003;133(5)(suppl 2):1592S-1596S. doi:10.1093/jn/133.5.1592S PubMedGoogle ScholarCrossref
    52.
    Nyamathi  AM, Leake  B, Gelberg  L.  Sheltered versus nonsheltered homeless women differences in health, behavior, victimization, and utilization of care.   J Gen Intern Med. 2000;15(8):565-572. doi:10.1046/j.1525-1497.2000.07007.x PubMedGoogle ScholarCrossref
    ×