Article Text

Residence change during the first trimester of pregnancy and adverse birth outcomes
1. Julia C Bond1,
2. Amanda L Mancenido1,
3. Divya M Patil1,
4. Seth S Rowley1,
5. Jack Goldberg1,
6. Alyson J Littman1,2,3
1. 1 Department of Epidemiology, University of Washington School of Public Health, Seattle, Washington, USA
2. 2 Seattle Epidemiologic Research and Information Center, Department of Veterans Affairs, Seattle, Washington, USA
3. 3 VA Puget Sound, Health Services Research and Development Center of Innovation for Veteran-Centered and Value-Driven Care, Seattle, WA, USA
1. Correspondence to Julia C Bond, Department of Epidemiology, University of Washington School of Public Health, Seattle, WA 98195, USA; jcbond{at}uw.edu

## Abstract

Background There are few published studies evaluating the impact of perinatal residence change on infant outcomes and whether these associations differ by socioeconomic status.

Methods We conducted a population-based cohort study using Washington State birth certificate data from 2007 to 2014 to assess whether women who moved during the first trimester of pregnancy (n=28 011) had a higher risk of low birth weight, preterm birth and small for gestational age than women who did not move during the first trimester (n=112 367). ‘Non-first-trimester movers’ were frequency matched 4:1 to movers by year. We used generalised linear models to calculate risk ratios and risk differences adjusted for maternal age, race, marital status, parity, education, smoking, income and insurance payer for the birth. We also stratified analyses by variables related to socioeconomic status to see whether associations differed across socioeconomic strata.

Results Moving in the first trimester was associated with an increased risk of low birth weight (6.4% vs 4.5%, adjusted risk ratio 1.37 (95% CI 1.29 to 1.45)) and preterm birth (9.1% vs 6.4%, adjusted risk ratio 1.42 (95% CI 1.36 to 1.49)) and a slight increased risk of small for gestational age (9.8% vs 8.7%, adjusted risk ratio 1.09 (95% CI 1.00 to 1.09)). Residence change was associated with low birth weight and preterm birth in all socioeconomic strata.

Conclusion Moving during the first trimester of pregnancy may be a risk factor for adverse birth outcomes in US women. Healthcare providers may want to consider screening for plans to move and offering support.

• low birth weight
• maternal stress
• moving
• pregnancy
• prenatal stress
• preterm birth
• residence change

## 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.

## Introduction

Moving is common in the USA. According to the US Census Bureau, 33.7% of Americans surveyed in 2015 had changed residences during the previous 5 years.1 Moving is especially common among young adults and is often precipitated by major life transitions such as having children.2

Environmental health cohort studies estimate that 11%–25% of women move residences while pregnant.3–5 Despite being a frequent occurrence, little is known about the potential health impacts of a move during pregnancy. There are compelling reasons to suspect that residence change during pregnancy may impact birth outcomes. A large body of literature links residential mobility during childhood with adverse outcomes in adolescence and adulthood, including drug abuse,6 lower educational achievement,7 suicidality8 and poor mental health,9 though the causal mechanisms of this relationship remain unclear.

Previous research suggests an association between maternal psychosocial stress and adverse birth outcomes, including preterm birth and low birth weight (LBW).10 11 Changing residences is often stressful, and thus could induce stress-related adverse health consequences.12–14 One prospective study of pregnant women found that greater exposure to everyday stressful life events (including residence change) was associated with an increased risk of preterm birth when experienced during the first trimester, but not the second or third.15 A survey of women in China found that exposure to stressful life events during the first and second trimesters was associated with LBW and preterm birth, with the association being strongest in the first trimester.16 A study evaluating maternal stress-related mechanisms found an increased risk for gastroschisis associated with residence change during pregnancy.17 No studies have explicitly evaluated the association between residence change and the more common birth outcomes of LBW, preterm birth and small for gestational age (SGA).

Our aim is to assess the association between residence change during the first trimester of pregnancy and LBW, preterm birth and SGA in a large, population-based cohort. We focused on the first trimester because previous studies have suggested that psychosocial stresses experienced during the first trimester more strongly impact birth outcomes compared with stressors experienced during the second or third trimester.16 18–20 Additionally, we stratified our analyses by socioeconomic variables to assess the possibility of effect modification as lower socioeconomic status may exacerbate move-related stress.12 The results of our study could have clinical implications for the many pregnant women who move in the USA.

## Methods

### Sample

We conducted a population-based retrospective cohort study using Washington State birth certificate data compiled by the Washington State Department of Health. Birth certificates include information on parental demographics, maternal reproductive history, prenatal care and infant birth information. Central to this analysis, the birth certificate includes length of time at current residence.

We included singleton births in Washington State between 2007 and 2014 to mothers 18 years and older. We excluded women younger than 18 years old due to the possibility of additional stresses associated with teen pregnancies as well as an increased risk of LBW and preterm birth associated with teenage pregnancy.21 Women who reported a residential address, length of time at residence and had a clinical estimate of gestational length were eligible for inclusion. A priori power calculations indicated that, assuming a prevalence of 5% for LBW,22 the outcome with the lowest prevalence, 80% power, and alpha=0.05, a sample size of 22 000 movers, matched 1:4 to women who did not move, would allow us to detect a relative risk (RR) ratio of 1.10 or greater. To account for the possibility of missing data we randomly selected 30 000 women who moved during their first trimester, frequency matched 1:4 by birth year with women who did not move in the first trimester for a total sample size of 150 000 women. Women without complete data on all prespecified confounders (n=1989 for first-trimester movers (6.6% of movers) and n=7549 for non-first-trimester movers (6.3% of non-movers)) were excluded from the analysis.

This research was considered exempt from review by the Washington State Institutional Review Board because all data were deidentified. All data protection regulations required by the Washington State Department of Health to ensure the integrity and anonymity of the data were implemented.

### Exposure to residential change

Moving in the first trimester was operationalised using gestational age and duration at current residence at birth. Women were classified as first-trimester movers if they changed residence during the first trimester of pregnancy and were considered non-first-trimester movers if they remained at the same residence during their pregnancy or changed residence during the second or third trimester of pregnancy. When gestational age at birth subtracted from duration at residence was greater than zero and less than or equal to 13 weeks, women were considered first-trimester movers. When gestational age at birth subtracted from duration at residence was less than zero or greater than 13 weeks, women were considered ‘non-first-trimester movers’. Women were excluded if estimated gestational age subtracted from reported duration at residence was exactly equal to zero due to the high likelihood of misclassification.

### Pregnancy outcomes

The outcomes of interest were LBW, preterm birth and SGA. Infants were classified as LBW if the weight on the birth certificate was <2500 g. Infants were classified as preterm if the gestational age based on clinical estimate was <37 weeks. As a supplementary analysis, we also examined very preterm birth (gestational age <32 weeks). SGA was defined as a birth weight in the lowest 10% of births for gestational age and sex.23

### Covariates

We used published literature to identify the correlates of residence change during pregnancy to inform our prespecified confounders, as well as how such factors were categorised.24–26 We considered maternal age (continuous), marital status (married, not married), parity (nulliparous, primiparous, multiparous), maternal race (Asian/Pacific Islander, Black, non-white Hispanic, Native American/Alaska Native, white) and smoking status (one or more cigarettes at any point during pregnancy, no smoking) as potential confounders. Higher residential mobility has been associated with lower socioeconomic status.12 24–26 We used median household income in the census tract of residence at time of birth (less than $40 000,$40 000–$49 999,$50 000–$59 999, over$60 000) and insurance payer for the birth (Medicaid, private/government/self-pay, other) as proxies of socioeconomic status since the birth records did not include a direct measure of household income. To assess whether associations were similar across socioeconomic strata, we conducted adjusted stratified analyses by these measures of socioeconomic status as well as maternal educational attainment (less than high school, completed high school or some college/associate’s degree, bachelor’s degree or more). Stratified analyses were adjusted for all potential confounding factors, including all other socioeconomic status variables. To evaluate whether there was evidence of increased care disruption among movers compared with non-movers, we used descriptive statistics to examine the unadjusted distribution of prenatal care according to the Kotelchuck Index across cohorts.27 The Kotelchuck Index was not included in adjusted analyses because it was possibly a mediating variable, and because missingness was high (8.5% for movers and 9.2% for non-movers).

We used generalised linear models with robust variance to estimate the crude and adjusted association between first-trimester moving and both LBW and preterm birth. Adjusted models were subsequently stratified by socioeconomic proxy variables. We report both RR and risk differences (unadjusted and adjusted) and 95% CIs. Adjusted models included all potential confounders. All analyses were completed using Stata V.14.2 (StataCorp, College Station, TX).

## Results

Our final sample included 28 011 first-trimester movers and 112 451 non-first-trimester movers randomly selected from the total pool of 708 529 births in Washington State between 2007 and 2014 (figure 1).

Figure 1

Sample selection flow chart.

Women who moved during their first trimester of pregnancy were more likely than women who did not move during their first trimester to be in their teens or 20s (69.5% vs 55.9%), to have not completed high school (15.5% vs 13.6%), to live in a census tract with a median household income less than \$40 000 (41.8% vs 35.8%) and to be nulliparous (49.3% vs 41.0%, table 1). Women who moved were also less likely to be married than non-first-trimester movers (61.1% vs 69.6%). Women who moved were more likely to have smoked during pregnancy (11.9% vs 8.9%). Movers were slightly more likely than non-movers to have inadequate prenatal care as measured by the Kotelchuck Index (15.7% vs 14.4%), but also more likely to report greater than adequate prenatal care (19.1% vs 18.0%). For crude associations between prespecified confounders and outcomes of interest, see online supplementary table 1.

### Supplemental material

Table 1

Baseline characteristics of complete case analysis of pregnant women who did or did not move during their first trimester of pregnancy in Washington State, 2007–2014

Moving in the first trimester was associated with an increased risk of both LBW and preterm birth (table 2). The prevalence of LBW among first-trimester movers was 6.4%, compared with 4.5% in the non-first-trimester movers group. The unadjusted RR of LBW among first-trimester movers compared with non-first-trimester movers was 1.43 (95% CI 1.36 to 1.51). This association remained statistically significantly elevated following adjustment for maternal age, race, marital status, parity, education, smoking, median income of the census tract and insurance payer for the birth (RR=1.37, 95% CI 1.29 to 1.45). Moving was associated with an excess of 14 LBW deliveries per 1000 births compared with the non-moving group (adjusted risk difference 1.42 per 100; 95% CI 1.11 to 1.71).

Table 2

Crude and adjusted associations between first-trimester moving and low birth weight and preterm births in complete cases

The prevalence of preterm birth among first-trimester movers was 9.1% compared with 6.4% among non-movers. The unadjusted RR of preterm birth among first-trimester movers was elevated compared with non-first-trimester movers (RR=1.42, 95% CI 1.36 to 1.49), and did not change appreciably following adjustment for confounders (RR=1.38, 95% CI 1.32 to 1.44). Moving was associated with an additional 24 preterm births per 1000 births compared with the non-moving group (adjusted risk difference 2.40 per 100; 95% CI 2.04 to 2.76). Results for very preterm birth were not appreciably different (online supplementary table 2).

The prevalence of SGA was slightly higher among movers (9.8%) than non-movers (8.7%), corresponding to an unadjusted association of 1.12 (95% CI 1.08 to 1.17). This association was not statistically significant after adjustment for confounders (adjusted RR 1.05, 95% CI 1.00 to 1.09).

In adjusted analyses stratified by socioeconomic status, associations between moving and risk of LBW and preterm birth remained significantly elevated for each level of maternal educational attainment, insurance payer for the birth and census-level median household income. Notably, the strength of the associations was slightly weaker among higher socioeconomic strata compared with lower socioeconomic strata defined by education and payer (tables 3 and 4).

Table 3

Adjusted relative risk of LBW, preterm birth and SGA among women who moved in the first trimester of pregnancy compared with those who did not move, stratified by socioeconomic status proxy variables

Table 4

Adjusted risk differences of LBW, preterm birth and SGA among women who moved in the first trimester of pregnancy compared with those who did not move, stratified by socioeconomic status proxy variables

## Discussion

This is the first population-based cohort study to evaluate residence change as an independent risk factor for birth outcomes. In women who gave birth in Washington State from 2007 to 2014, moving residences during the first trimester was associated with an increased risk of LBW and preterm birth, but not SGA, as compared with women who did not move in the first trimester. The increased risk of LBW and preterm birth persisted after controlling for potential confounders and was present among women across both lower and higher socioeconomic status markers.

Previous studies have taken advantage of natural experiments to examine the effect of acute stressors such as natural disasters or national economic crises during pregnancy.19 20 28 29 Many of these studies found that the association between stress and LBW or preterm birth was strongest in the first trimester, compared with the second or third.19 20 28 Other studies focusing on exposure to life stressors (eg, bereavement, job loss) have also found stronger associations with adverse birth outcomes when stress occurred in the first trimester.15 16 Our study is consistent with the hypothesis that exposure to a stressful event in the first trimester increases the risk of adverse birth outcomes. Changing residences can be comparatively less stressful than a natural disaster, so it is noteworthy that our results were consistent with these studies.

Multiple plausible pathways by which residential mobility could impact birth outcomes exist: inability to obtain regular healthcare; physical strain associated with moving30; disruptions to social support, which may provide a buffer against deleterious effects of stress31; and/or a biological stress pathway triggered by the psychosocial stress of moving itself or a stressful situation precipitating the move. These are not mutually exclusive possibilities, and our ability to evaluate them using birth certificate data is limited.

Previous research has suggested that there may be different aetiological mechanisms underlying LBW for SGA and non-SGA babies.32 Because we did not see a strong association between residence change and SGA, it is possible that our observed associations between residence change and LBW are driven primarily by earlier births, as opposed to growth restriction. A cohort study in Japan found that women with delayed prenatal care initiation had infants with an increased likelihood of LBW in the absence of SGA.32 They hypothesised that delayed prenatal care may adversely influence outcomes. We assessed differences in prenatal care adequacy using the Kotelchuck Index on the birth certificate and saw only small differences between movers and non-movers, though this analysis was limited because approximately 9% of the sample had missing data on prenatal care. It is also possible that the Kotelchuck Index is not sufficiently detailed to capture disruptions to prenatal care due to residence change, which may be transient and could also be emotional in nature, such as discomfort with a new care team.

Though the large sample size of our study was a strength, some features of our study design limited the interpretation of our findings. Women reported how long they had been at their current residence in years or months on their birth certificate, which means that there could be some imprecision in terms of identifying when women moved in relation to when they conceived. We believe that this misclassification would likely be non-differential, however, and would bias our results towards the null. We were also unable to ascertain why women changed residences. Previous studies have found that women who change residences during pregnancy, particularly those who change residences frequently, are more likely to be of lower socioeconomic status than women who do not.12 24–26 In our study, movers tended to be younger, more likely to have a birth paid for by Medicaid and more likely to be living in a census tract with a lower median income than non-movers at the time of the birth; these are all factors associated with lower socioeconomic status. Moving during pregnancy could indicate a desire to leave risk factors such as environmental exposures or unstable living situations that could themselves impact birth outcomes. It is possible that the reasons for a move, including whether the move is voluntary or forced, differentially impact outcomes. Our data only contain the most recent residential address. Thus, we cannot know if women are moving into more or less desirable locations, which may also be relevant. Our observed associations could represent an average of heterogeneous effects depending on the motivation for the residential move.

Our stratified analyses suggest that the association between moving residences is associated with LBW and preterm birth for women across levels of socioeconomic status, though there was a slightly stronger association among women with markers of lower socioeconomic status. This may mean that the stressors associated with a residential move, which can include disruption of social support,33 impact women regardless of their socioeconomic resources. It is also possible that our socioeconomic proxy variables were not able to adequately explore the strength and nature of the relationship, particularly since we did not have a person-level measure of income. Future research should collect more detailed information on socioeconomic status to enable a more detailed assessment of this relationship.

Additionally, our analysis was restricted to live births. We cannot make any inferences about the association between residential change and pregnancy loss or stillbirth. According to the Washington State Department of Health, the rate of fetal mortality (defined as a gestation of 20 weeks or more not resulting in a live birth) was 6.0 per 1000 live births.22 As a result, it is possible that our study underestimates the impacts of residence change if there are impacts on fetal loss as well.

It is possible that exposure to stressful events in the second or third trimester may result in adverse outcomes. However, analyses using our data set to assess the impact of second and third-trimester moving would have been biased because of the conditioning on longer pregnancies (ie, all women would need to maintain the pregnancy at least until the second trimester to have the potential to move during the second trimester). Future studies evaluating the impact of moving by trimester should consider a case–control or matched design to account for the bias inherent in assessing exposures that occur later in pregnancy.

Despite these limitations, our results yield important insights regarding moving during pregnancy. The large, population-based cohort from Washington State suggests generalisability of our findings. The associations between residence change and adverse infant outcomes were present across all levels of socioeconomic status. Thus, moving may be a useful marker for care providers of pregnant women. Regardless of whether the negative impact of moving is driven by the stress from the move itself, stressful situations leading to a move, or disruption of care because of the move, asking patients about plans to move and using that as an opportunity to counsel patients on stress-mitigating techniques and care continuity may be beneficial. Further, there is some evidence suggesting that maternal perception of stress may be a mediating factor in the association between stressful events and birth outcomes.15 34 35 Though residence change during pregnancy is common in the USA, clinicians may wish to more carefully monitor particularly stress-prone patients if they will be changing residences during their first trimester.

Future research should include both qualitative and quantitative efforts to evaluate maternal residence change as an independent risk factor for birth outcomes and should assess both the reasons for the move and how stressful the move was for women. Because up to 25% of pregnant women are estimated to move during pregnancy, elucidating associations between moving and adverse birth outcomes has the potential to impact the clinical care of many women and have an important impact on public health.3 25

### What is already known on this subject

• Maternal psychosocial stress due to life events can adversely impact infant outcomes. A large fraction of women change residences during their childbearing years. Residence change during childhood has been associated with adverse health outcomes for children, but little is known about the impact of residence change during pregnancy on infant outcomes.

• This is the first large, population-based study to evaluate residence change as an independent risk factor for infant outcomes. We found that moving during the first trimester of pregnancy was associated with an increased risk of preterm birth and low birth weight. These associations persisted within socioeconomic strata, though with slightly stronger associations in those with lower socioeconomic status. Our results suggest that residence change may be associated with adverse infant outcomes.

## Footnotes

• Contributors All authors contributed to the conceptual development of the research, review of analysis and interpretation of results. Data acquisition was completed by SSR. The analytical work was conducted primarily by JCB with assistance from ALM and supervision by AJL. JCB drafted the article with assistance from AJL. All authors critically revised the article and approved the final version for publication.

• Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

• Competing interests None declared.

• Patient consent for publication Not required.

• Ethics approval Washington State Institutional Review Board Project E-021617.

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

• Data availability statement Data may be obtained from a third party and are not publicly available.