key: cord-0919485-258l3jrk
authors: FISHER, Stephanie A.; SAKOWICZ, Allie; BARNARD, Cynthia; KIDDER, Seth; MILLER, Emily S.
title: Neighborhood deprivation and preterm delivery during the coronavirus 2019 pandemic
date: 2021-09-22
journal: Am J Obstet Gynecol MFM
DOI: 10.1016/j.ajogmf.2021.100493
sha: fbd461c0456979801e39d5069088fd667891b269
doc_id: 919485
cord_uid: 258l3jrk

Background Prior studies have reported decreases in preterm delivery (PTD) incidence during the coronavirus 2019 (COVID-19) pandemic, however findings are inconsistent. Given wide disparities in the pandemic's impact across communities, neighborhood deprivation may explain observed variation in the relationship between the COVID-19 pandemic and preterm delivery. Objective To characterize changes in the incidence of PTD during the COVID-19 pandemic with attention to effect modification introduced by neighborhood hardship. Study Design This retrospective cohort study included all pregnant patients who delivered at an urban tertiary care hospital during the pandemic (April-November 2020) or pre-pandemic (April-November 2019). We compared the incidence of PTD, spontaneous PTD, and medically indicated PTD prior to 37 weeks’ gestation across epochs. Planned analyses stratified the cohorts by neighborhood deprivation metrics defined by residential zip code, including median neighborhood household income and hardship index (a composite index including dependency, educational attainment, unemployment, poverty, per capita income, and crowded housing). The Breslow-Day test for homogeneity assessed the association of delivery epoch and neighborhood deprivation with PTD outcome. Results Of 16,544 eligible deliveries, 8.7% occurred preterm. Incidences of PTD (8.4% vs. 9.0%, p=0.17), spontaneous PTD (5.0 vs 5.4%, p=0.27), and medically indicated PTD (3.2% v 3.5%, p=0.47) were similar between the pandemic and pre-pandemic epochs. However, PTD (OR 0.78, 95% CI 0.64-0.96) and spontaneous PTD (OR 0.76, 95% CI 0.59-0.99) decreased from the pre-pandemic to pandemic epoch in those living in neighborhoods <50th percentile for median income (Breslow-Day p-values 0.047 and 0.036, respectively). Similarly, PTD (OR 0.78, 95% CI 0.64-0.97) and spontaneous PTD (OR 0.74, 95% CI 0.57-0.98) decreased for those inhabiting neighborhoods in the highest-hardship quartile (Breslow-Day p-values 0.045 and 0.029, respectively). Conclusion Populations residing in socioeconomically disadvantaged neighborhoods experienced reductions in preterm delivery during the COVID-19 pandemic. Neighborhood-level social determinants of health offer insight into the complex etiologies that contribute to preterm delivery, and provide opportunities for public health, equity-focused prevention strategies.

The World Health Organization declared the novel coronavirus 2019 (COVID-19) a global pandemic on March 12, 2020. 1 As the United States entered lockdown, the impact was not distributed equally. In particular, Chicago, the third-most populous city in the United States composed of 77 distinct community areas with a wide breadth of social and economic diversity, has experienced significant differential neighborhood-level impact of the pandemic. [2] [3] [4] [5] Income level strongly influences an individual's ability to take protective measures against viral transmission, including staying home. 6 Essential workers disproportionately reside in higher hardship communities and have had less ability to work from home; however, service and low-income workers may also have been more likely to lose their employment as many small businesses and restaurants closed or reduced staff, forcing them to stay home. [6] [7] [8] Just as shutdown measures appear to have averted influenza season altogether, it is possible that forced reductions in discretionary activity and an increased proportion of community members quarantining at home during the pandemic may have had protective effects on pregnant patients normally exposed to a range of risks in normal occupational and social life. [6] [7] [8] Given the increased adverse obstetric outcomes associated with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection in pregnancy, including increased preterm delivery (PTD) rates, pregnant workers may have preferred to limit exposures during the pandemic, accepting forced unemployment and lockdown at home. [9] [10] [11] One-third of pregnant patients stayed home from work due to fear of contracting the virus, and almost half altered their primary mode of transportation. 11 These and other pandemic-related maternal behavior changes may influence obstetric outcomes. 12 Several reports on the incidence of preterm delivery (PTD) during the pandemic have emerged from medical centers in Europe, Asia, and northeastern North America with mixed results. While some studies found a decreased incidence of preterm delivery 1,13-17 , others have not identified any significant change. [18] [19] [20] [21] Additionally, differences in preterm delivery rates according to race, ethnicity, or insurance payer type have not been identified. 19, 21, 22 However, these studies have not evaluated the extent to which other social determinants of health (SDOH) may have modified their findings. Insofar as SDOH may be imputed from neighborhood hardship and is associated with adverse perinatal outcomes, including PTD 23 , understanding the interplay between neighborhood environment and observed changes in PTD may explain the observed variation. Thus, we aimed to characterize changes in PTD incidence during the COVID-19 pandemic with particular attention to effect modification introduced by neighborhood hardship. We hypothesized that PTD incidence would be similar overall during the pandemic as compared to pre-pandemic, but the incidence of PTD would be reduced during the pandemic specifically in areas of high neighborhood hardship.

This retrospective cohort study was conducted at Northwestern Memorial Hospital's Prentice Women's Hospital, an urban academic tertiary care center in Chicago, IL. Pregnant patients who delivered from April 1, 2019 to November 30, 2019 represented the pre-pandemic epoch; those who delivered from April 1, 2020 to November 30, 2020 represented the pandemic epoch. Similar periods during the pandemic and pre-pandemic months were utilized to account for seasonal changes in delivery volume at our center, which has a delivery volume of approximately 12,000 patients per year. At our institution, historically the majority of deliveries of patients cared for in our system are inborn and no secular changes were noted during the pandemic with respect to inborn deliveries.

We included pregnant patients age 18 years or older, and excluded individuals who underwent pregnancy termination or delivered before 20 weeks' gestation. Eligible patients were identified via query of the electronic health record. Sociodemographic characteristics, including maternal age, self-reported race and ethnicity, preferred language, and insurance status were abstracted. We also abstracted obstetric characteristics, including parity, any prior history of PTD, and presence of a multiple gestation. We confirmed pregnancy dating and gestational age at delivery by last menstrual period and the earliest ultrasound performed. 24 We also identified patients with a SARS-CoV-2 diagnosis during the pregnancy who delivered during the pandemic, as all patients underwent universal screening on admission for labor and delivery after April 2020.

Recorded zip code of residence was used to delineate levels of neighborhood deprivation.

Specifically, we identified median neighborhood annual household income and dichotomized each patient by whether the zip code was at or above versus below the median annual household income for Chicago, which was $58,247 in 2019. 25 In addition, we assigned a combined hardship index score to each patient according to residential zip code from the 2015-2019 American Community Survey 5-year estimates. 26, 27 This index score averages six variables on a standardized scale: dependency (population <18 or >64 years of age), education (percent of residents over age 25 without a high school diploma), unemployment (percent of residents over age 16 and unemployed), poverty (percent of households below the federal poverty line), per capita income level, and crowded housing (percent of housing units with more than one person per room). Higher hardship index scores indicate a greater degree of socioeconomic disadvantage and social vulnerability. The calculated hardship index was divided into quartiles, and the highest-hardship quartile was compared to the lower three quartiles. A similar approach was used to evaluate specific subcomponents of the hardship index.

The primary outcome was PTD, defined as any delivery prior to 37 weeks' gestation, which includes stillbirths as these are not mutually exclusive events. Secondary outcomes included spontaneous and medically indicated PTD, which did not include stillbirths. To ascribe an etiology (spontaneous or medically indicated) to each PTD, the medical record of each PTD was abstracted by a Maternal-Fetal Medicine clinician (SAF) who was blinded to the epoch.

Spontaneous PTDs included those that occurred in the setting of preterm labor, preterm premature rupture of membranes, or placental abruption. Medical indications for indicated PTDs included pre-eclampsia or other hypertensive disorder of pregnancy, fetal indication, or abnormal placentation.

We compared maternal baseline sociodemographic, obstetric, and neighborhood deprivation characteristics by epoch with descriptive characteristics and frequencies reported, and bivariable analyses performed. Independent samples t-tests compared normally distributed continuous variables. Chi-square tests compared categorical variables. We performed generalized estimating equation log-binomial regression for binary outcomes for the primary and secondary outcomes to compare odds ratios during the pandemic and pre-pandemic epochs, accounting for the clustering effect of multiple gestations. Odds ratios were adjusted for those baseline characteristics found to differ across epochs (p<0.05) in bivariable analysis.

We performed generalized estimating equation log-binomial regression subgroup analysis for spontaneous and medically indicated PTD stratified by neighborhood deprivation metrics, including median annual household income percentile by zip code (<50 th percentile versus >50 th percentile), and hardship index (highest-hardship quartile versus lower three quartiles) for both the combined hardship index and its subcomponents. We utilized the highest-hardship quartile in an attempt to differentiate those experiencing the highest degree of hardship, as we hypothesized that those with a more extreme exposure to social determinants of health would exhibit the most significant changes in PTD incidence. Analyses using the Breslow-Day test of homogeneity were planned for PTD categories that demonstrated significant differences in bivariable subgroup analysis to determine whether effect modification was present across neighborhood deprivation metrics. A p-value <0.05 was considered statistically significant.

Statistical analyses were conducted using Stata version 16.1 (College Station, TX). The study was approved by the Northwestern University Institutional Review Board prior to its initiation with a waiver of informed consent.

During the study period, the analytic cohort encompassed 16,544 deliveries ( Figure 1 ) with baseline sociodemographic characteristics described in Table 1 . Small, but statistically significant, differences were noted for self-reported race and prevalence of multiple gestations across epochs. Notably, no differences were noted in neighborhood deprivation across epochs.

During the pandemic epoch, 5.9% of patients had SARS-CoV-2 infection during their pregnancy.

A total of 1,435 deliveries (8.7%) occurred preterm, 694 (48.4%) of which occurred during the pandemic epoch. We did not detect a significant difference in the odds of PTD, spontaneous PTD, or medically indicated PTD across epochs in bivariable analyses or after adjusting for race and multiple gestations in multivariable analyses ( Table 2 ). As self-reported race and ethnicity represent social constructs without a biologic basis for association with PTD, additional sensitivity analysis excluding race and ethnicity from multivariable analysis was performed without a change in these findings.

We then evaluated the odds of PTD, spontaneous PTD, and medically indicated PTD during the pandemic and pre-pandemic epochs according to neighborhood deprivation metrics.

Among patients who reside in zip codes with median household income >50 th percentile, the incidences of PTD overall, spontaneous PTD, and medically indicated PTD were similar in the pandemic and pre-pandemic epochs (Table 3 ). However, among pregnant patients who reside in zip codes with median annual household income <50 th percentile, the incidences of PTD overall (absolute risk reduction [ARR] -2.3%, p=0.02) and spontaneous PTD (ARR -1.6%, p=0.03) were significantly decreased during the pandemic compared to the pre-pandemic epoch. The incidence of medically indicated PTD did not differ between epochs among pregnant patients who reside in lower income areas.

Among pregnant patients in the highest-hardship quartile for hardship index, incidence of PTD overall (ARR -2.4%, p=0.02) and spontaneous PTD (ARR -1.8%, p=0.03) was significantly lower during the pandemic compared to the pre-pandemic epoch. Within the individual hardship index subcomponents, education and crowded housing indices contributed to significant reductions in PTD overall and spontaneous PTD observed during the pandemic among the highest-hardship quartile (Table 4 ). While we did not detect statistical significance for dependency, education, poverty, and per capita income, point estimates of PTD were all decreased in areas of greater hardship during the pandemic epoch, irrespective of metric used.

This pattern did not hold true for patients in the lower three hardship quartiles, where the incidences of PTD overall and spontaneous PTD were similar between epochs ( Table 3 ). The incidence of medically indicated PTD did not differ significantly among pregnant patients according to hardship index quartile, including for the combined hardship index and its individual components.

We then performed the Breslow-Day test of homogeneity to compare PTD overall and spontaneous PTD between the pandemic and pre-pandemic epochs according to income percentile by zip code and hardship indices ( Table 5 ). The observed differences in the odds of PTD and spontaneous PTD between epochs differed significantly for income percentile by zip code and the combined, education, and crowded housing indices, further indicating a significant association between neighborhood hardship and changes in PTD incidence during the pandemic.

This study of a large, diverse cohort found PTD incidence was similar during the COVID-19 pandemic compared to a similar pre-pandemic epoch. However, pregnant patients residing in socioeconomically disadvantaged neighborhoods had lower rates of PTD and spontaneous PTD during the pandemic. In contrast to our findings, a large national registry in the Netherlands suggested that reductions in PTD were confined to people living in highsocioeconomic status neighborhoods. 15 Domestically, a study evaluating pregnant patients in Pennsylvania demonstrated decreases in spontaneous PTD rates were limited to women living in more advantaged neighborhoods determined by area deprivation indices. 28 To the best of our knowledge, this is the first study to specifically link increased neighborhood hardship with decreased PTD and spontaneous PTD rates during the pandemic.

Our finding of a lower rate of PTD overall and spontaneous PTD during the pandemic among patients residing in lower income areas with greater economic hardship may explain conflicting reports published previously regarding the effect of the COVID-19 pandemic on PTD rates. These prior evaluations did not account for neighborhood-level SDOH. By using the combined economic hardship index as well as other individual markers of economic hardship, we were able to identify specific economic factors which may be associated with a lower rate of PTD due to spontaneous PTD in socially vulnerable patients.

The etiology of spontaneous PTD is likely multifactorial and largely unknown. 20 Race, ethnicity, and lower income level have been associated with increased PTD rates domestically and globally. [29] [30] [31] However, race and ethnicity have not been significantly associated with changes in PTD during the pandemic 21, 22 , and our findings suggest other SDOHs, particularly an individual's neighborhood environment, more significantly impact PTD risk. Individual indices for education and crowded housing within the hardship index were key contributors to the reduction in PTD overall and spontaneous PTD observed during the pandemic. While not statistically significant, all other indices of neighborhood hardship exhibited consistently lower point estimates for PTD and spontaneous PTD incidence during the pandemic.

Prior studies have associated lower maternal educational attainment and housing instability with increased incidence of PTD. [32] [33] [34] The seemingly contradictory reduction in PTD we identified among this subgroup during the pandemic may relate to unique pandemic-driven changes that occurred in the economy and workforce. The pandemic disproportionately affected low-wage workers and individuals from lower-income backgrounds. 35 Between February-May 2020, the unemployment rate for workers with a high school diploma or less rose by more than higher. 36 Maternal stress, shift work and other work-related stressors, and working ≥40 hours per week have specifically been associated with increased PTD risk more than five-fold. 37, 38 Data also demonstrate that individuals with less educational attainment are more likely to work in a job with medium-or high-contact requiring greater public interaction than workers with a bachelor's degree or higher. 39 Taken together, these statistics support the theory that pregnant individuals with a lower educational attainment were more likely to become unemployed during the pandemic, resulting in decreased public interaction and work-related stress, and ultimately leading to a decreased incidence of PTD. These changes, in turn, may have modulated the immune system response that plays an active role in preterm labor initiation. [40] [41] [42] With additional consideration to unemployment, the financial stress associated with loss of income for those unemployed during the pandemic was partially offset by the unprecedented income support and eviction moratoria by the federal government as part of the pandemic response. As a result, pandemic-associated unemployment was, for some, a less financially stressful experience than unemployment might have been under standard circumstances. The pandemic may further have impacted the nature of social support available for pregnant people.

Low or inconsistent social support has a negative impact on pregnancy-related outcomes. 43 In light of the stressful and uncertain pandemic circumstances, the increased time at home with partners and family members within a defined "quarantine bubble" may have offered opportunities for greater social support, further reduction in stress and resultant immunomodulation, and positive effects on overall well-being. 44 It is possible that these changes in maternal behavior and increased social and financial support for lower-income pregnant patients improved obstetric outcomes, including reduced rates of spontaneous PTD.

Pregnant patients from lower-income areas may also face systems-level barriers to accessing prenatal care, such as transportation difficulties and an inability to take time off of work. 45, 46 Additional barriers may have presented themselves during the pandemic, including fear of viral transmission related to seeking medical care and additional childcare responsibilities as a result of school and daycare closures. Increased telemedicine utilization during the pandemic, including at our institution, has led to broader accessibility to high-quality prenatal care for many pregnant patients, albeit remote care. Consistent prenatal care is associated with a decreased incidence of PTD. 47 While the implementation and uptake of telehealth during the pandemic has likely been variable among different socioeconomic groups in various regions of the country, the potential for improved access to care for lower-income pregnant patients, who may have faced increased barriers to care prior to the widespread adoption of telemedicine during the pandemic, could have contributed to decreased spontaneous PTD rates in lower-income patients. In order to attribute a net protective effect, it would be hypothesized that these potential benefits of pandemic-driven home and occupational changes outweighed the known higher risks of PTD related to income and other SDOH.

Future research to provide more specific insight into changes in prenatal care delivery via telemedicine and alterations in workforce, employment, and work-from-home policies in the highest hardship neighborhoods during the COVID-19 pandemic is indicated to better understand their association with decreased PTD rates. Evaluation of how government policies regarding employment, wage, and housing assistance and access to transportation and technology as part of the pandemic response have impacted pregnancy outcomes has the potential to drive future public policy. Following recovery from the pandemic, further evaluation of public health interventions to address socioeconomic disparities and the resultant impact on PTD in disadvantaged communities is warranted. Finally, future studies should similarly explore the association of socioeconomic status on differential rates of other obstetrical adverse pregnancy outcomes, beyond preterm delivery, during the pandemic and pre-pandemic epochs, such as the incidence of hypertensive disorders of pregnancy and perinatal mood disorders that may be associated with increased stress and hardship during the COVID-19 pandemic.

A major strength of this study is the large, diverse sample size encompassing epochs before and during the pandemic. This data stemmed from a comprehensive clinical data set of pregnant people who delivered during specified time epochs, limiting the potential for selection bias. This study is also strengthened by our use of consistent pregnancy dating methods and manual chart review to systematically assess PTD indications (spontaneous versus iatrogenic), which may not be feasible in database or national registry studies examining rates of PTD during the pandemic. Finally, use of the American Community Survey's multi-year estimates facilitated increased statistical reliability of economic hardship data, especially for less populated areas and small population subgroups.

This study is also subject to limitations. These data are limited to a single tertiary care institution and our results may not be generalizable to other centers. The hardship index's unemployment subcomponent was derived prior to the pandemic and may not fully reflect changes in employment status and hardship that occurred during the pandemic. We also recognize that we imputed hardship based on zip code, and this is only one marker of hardship which may not represent the socioeconomic characteristics of individual persons within the community. Area level analysis by zip codes is further limited by their large, variable sizes and heterogeneity of SDOH across these areas; however, zip code has been shown to be an important marker to understand the overall needs of a population, and our use of data from the American Community Survey allowed us to incorporate several measures of economic hardship. Finally, our study is limited by inability to adjust for unmeasured confounders and patient-level factors as a result of our observational and retrospective study design, which may influence the observed findings. While many interesting hypotheses may be generated from the results of our study, our conclusions regarding the observed associations must remain tempered and cannot suggest causation.

SDOH are established contributors to health outcomes, and PTD is no exception. The unique intersectionality between neighborhood-level social determinants and changes resulting from the COVID-19 pandemic may have an impact on adverse pregnancy outcomes, such as PTD. These findings may offer greater insight into the complex etiologies that contribute to PTD and public health and equity-focused strategies for prevention. 

Danish premature birth rates during the COVID-19 lockdown

A Kaleidoscope of Culture: Measuring Diversity in Chicago's Neighborhoods

DePaul University

Social Vulnerability and Racial Inequality in COVID-19 Deaths in Chicago

Racial/ethnic minority and neighborhood disadvantage leads to disproportionate mortality burden and years of potential life lost due to COVID-19 in Chicago

Coronavirus disease 2019 (COVID-19) mortality and neighborhood characteristics in Chicago

Neighbourhood income and physical distancing during the COVID-19 pandemic in the United States

Metropolitan Chicago's essential workers disproportionately low-income, people of color

Chicago Metropolitan Agency for Planning

Labor Force Statistics from the Current Population Survey

Birth and infant outcomes following laboratory-confirmed SARS-CoV-2 Infection in Pregnancy -SET-NET, 16 Jurisdictions

Outcome of coronavirus spectrum infections (SARS, MERS, COVID-19) during pregnancy: a systematic review and meta-analysis

Health anxiety and behavioural changes of pregnant women during the COVID-19 pandemic

Effect of the COVID-19 pandemic response on intrapartum care, stillbirth, and neonatal mortality outcomes in Nepal: a prospective observational study

Unprecedented reduction in births of very low birthweight (VLBW) and extremely low birthweight (ELBW) infants during the COVID-19 lockdown in Ireland: a 'natural experiment' allowing analysis of data from the prior two decades

Calgary doctor says' dramatic' 40% drop in PTDs during lockdown could be global trend

Impact of COVID-19 mitigation measures on the incidence of PTD: a national quasiexperimental study

Decreased incidence of PTD during coronavirus disease 2019 pandemic

Change in the incidence of stillbirth and preterm delivery during the COVID-19 pandemic

PTD During the Coronavirus Disease 2019 (COVID-19) Pandemic in a Large Hospital System in the United States

Have Coronavirus Disease 2019 (COVID-19) Community Lockdowns Reduced PTD Rates?

Singleton preterm birth rates for racial and ethnic groups during the coronavirus 2019 pandemic in California

Racial/Ethnic Disparities in Very Preterm Birth and Preterm Birth Before and During the COVID-19 Pandemic

Area-level deprivation and preterm birth: results from a national, commercially-insured population

Committee Opinion No 700: Methods for Estimating the Due Date

Understanding Central City Hardship

American Community Survey 5-Year Data

What is driving the decreased incidence of preterm birth during the coronavirus 2019 pandemic?

Socio-economic disparities in pregnancy outcome: why do the poor fare so poorly?

Birth outcomes and infant mortality by income in urban Canada

Low birth weight in Spain associated with sociodemographic factors

Mother's education and the risk of preterm and small for gestational age birth: a DRIVERS meta

Mother's education and the risk of several neonatal outcomes: an evidence from an Italian population-based study

Associations between unstable housing, obstetric outcomes, and perinatal health care utilization

About Half of Lower-Income Americans Report Household Job or Wage Loss Due to COVID-19. Pew Research Center's Social & Demographic Trends Project

COVID-19, educational attainment, and the impact on American workers

The preterm prediction study: maternal stress is associated with spontaneous preterm birth at less than thirtyfive weeks' gestation. National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network

Lifeways Cross-Generation Cohort Study Steering Group. Occupational predictors of pregnancy outcomes in Irish working women in the Lifeways cohort

A Basic Demographic Profile of Workers in Frontline Industries

Prenatal Maternal Stress Causes Preterm Birth and Affects Neonatal Adaptive Immunity in Mice. Front Immunol

Pathogenesis of preterm birth: bidirectional inflammation in mother and fetus

The contribution of maternal stress to preterm birth: issues and considerations

Barriers and Facilitators to Recurrent Preterm Birth Prevention among Low-Income Women: A Qualitative Study

Perceived partner responsiveness as an organizing construct in the study of intimacy and closeness

Anatomy of Good Prenatal Care: Perspectives of Low Income African-American Women on Barriers and Facilitators to Prenatal Care