Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Mar 7.
Published in final edited form as: Demography. 2025 Aug 1;62(4):1155–1169. doi: 10.1215/00703370-12178940

Dynamic Family Size Preferences During the COVID-19 Mortality Crisis: A Research Note

Letícia J Marteleto 1, Sneha Kumar 2
PMCID: PMC12965107  NIHMSID: NIHMS2149790  PMID: 40771040

Abstract

In this research note, we examine how family size preferences evolved for women with and without children in response to changing COVID-19 mortality exposure during the first two years of the COVID-19 pandemic. We leverage spatiotemporal variation in COVID-19 deaths occurring during panel surveys in 2020 and 2021 with a population-based sample of 2,520 women, aged 18–34, across 94 municipalities in Pernambuco, Brazil. We use individual fixed-effects regressions to examine whether changes in municipality-level COVID-19 death rates are associated with changes in women’s desired family size, net of their own or their family’s COVID-19 infection status and other time-varying sociodemographic factors. We find that women with and without children at baseline responded differently to changing municipality-level COVID-19 deaths: while women without children did not change their desired family size, women with children saw a small but significant increase in their desired family size in response to rising COVID-19 mortality. These innovative findings suggest that women with children responded to widespread COVID-19-related loss within their communities by wanting to build and consolidate their families. We advance knowledge about varying contextual influences on fertility preferences during epidemics in a middle-income country with early and below-replacement fertility.

Keywords: Fertility preferences, Desired family size, COVID-19, Mortality shocks, Latin America

Introduction

This research note examines how women’s family size preferences evolved in response to changing exposure to COVID-19 mortality during the first two years of the pandemic. Recent evidence has shown how the pandemic affected and will continue to affect fertility preferences (Lazzari et al. 2024; Lindberg et al. 2020; Luppi et al. 2020; Manning et al. 2022; Zimmerman et al. 2023; Zimmerman et al. 2022) and birth rates (Aassve et al. 2021; Sobotka et al. 2021). Yet, there is no evidence on whether varying exposure to COVID-19 mortality and its associated uncertainties translated into meaningful variation in family size preferences. Specifically, it remains unclear whether changing local-level COVID-19 mortality over time resulted in any upward or downward adjustments in desired family size (DFS) and, if so, whether there is variation by life course stage.

We draw on a population-based panel of women aged 18–34 who were interviewed in 2020 and 2021 across 94 municipalities in Pernambuco, Brazil, to examine how DFS changed in response to changing exposure to COVID-19 mortality. We leverage two unique study design features for this analysis.

First, in contrast to research based on other settings with little or no contextual variation in mortality experiences or risk, we leverage spatiotemporal variation in mortality occurring during our panel surveys to examine DFS. Understanding how macrostructural crises in general and epidemics in particular affect fertility preferences requires going beyond binary measures of whether people lived through the “threat” period. Instead, we need to account for their degree of exposure to the “threat,” which can vary across geographic areas (Agadjanian and Prata 2002) and over time (Heuveline and Poch 2007; Sobotka et al. 2011). When a novel infectious disease crisis such as COVID-19 emerges, the degree of exposure to mortality at the contextual level is especially relevant—it signifies the extent of life loss, immediate mortality risk, socioeconomic destruction, and unpredictability people observe.

Second, unlike other studies on COVID-19 and on other crises in general, which are often unable to disentangle overlapping dimensions of a crisis, we account for the multidimensional nature of the pandemic. We control for co-occurring changes in women’s socioeconomic and life course conditions, including changes in one’s own or one’s family’s COVID-19 infection status, to adequately isolate how widespread loss and uncertainty associated with rising local-level COVID-19 mortality contributed to changing DFS.

To our knowledge, this is the first study to underscore how a distinctive element of the pandemic—varying local-level COVID-19 deaths—informed DFS trends for women in different life stages. While existing studies on COVID-19 (Koenig et al. 2022; Lazzari et al. 2024; Manning et al. 2022; Rocca et al. 2022; Zimmerman et al. 2022) and other public health crises (Trinitapoli and Yeatman 2011; Yeatman 2009) have highlighted individual-level correlates of changing fertility preferences, our study underscores the significance of changing local epidemiological conditions while simultaneously accounting for individual-level infection and other concurrent changes. We also look at a different stage of the pandemic than other COVID-19 studies that limit analyses to 2020 and other studies that have focused on long-term effects of a public health crisis but not its initial stages.

Our focus on Brazil also represents an important contribution. Brazil is characterized by “pervasive socioeconomic inequality, high levels of informality in the labor market, weak social protection systems, and slow progress toward gender equality” (Esteve et al. 2022:485). This made young Brazilians (unequally) susceptible to volatile DFS during the pandemic, as the country faced some of the highest COVID-19 death tolls globally—including increased maternal and infant mortality. This mortality context is also noteworthy given Brazil’s recent exposure to an epidemic centered on fetal health risks: Zika in 2015–2017. While fertility has been below replacement since the early 2000s, there were some birth rate declines in the aftermath of Zika (Marteleto et al. 2020). Subsequently, the total fertility rate in Brazil and Pernambuco went from 1.66 and 1.70 in 2020, respectively, to 1.64 and 1.68 in 2021 (Instituto Brasileiro de Geografía e Estatística [IBGE] 2024). This analysis, therefore, offers critical insight into whether COVID-19 deaths quelled fertility desires and potentially limited parity progression or whether they generated pronatalist tendencies.

Notably, this is the first study to examine fertility preferences and DFS in a dynamic way in Latin America, regardless of COVID-19. Apart from a few exceptions in African countries (Bankole and Westoff 1998; Müller et al. 2022; Sennott and Yeatman 2012; Trinitapoli and Yeatman 2018; Yeatman et al. 2013), most studies investigating dynamic individual-level fertility preferences have focused on North America and Western Europe (e.g., Hayford 2009; Heiland et al. 2008; Iacovou and Tavares 2011; Miller et al. 2013; Rocca et al. 2010).

Hypotheses

There are competing hypotheses for how DFS could have changed in response to varying COVID-19 mortality, though none have been tested yet. We draw on studies documenting fertility responses to various mortality shocks for the theoretical framework guiding our expectations, while recognizing that each mortality shock is unique with distinct possible pathways of influence on fertility.

One hypothesis is that women may have reduced their DFS with rising COVID-19 mortality, as these deaths may have increased worries around childbearing. COVID-19 infection during pregnancy is linked to problems such as preeclampsia and pre-term birth (Celewicz et al. 2023; Edlow et al. 2022), and perceived risks of these problems would be especially relevant in Brazil, which saw some of the world’s highest COVID-19-related increases in maternal and postpartum mortality (Diniz et al. 2022; Gonçalves et al. 2021), as well as infant and child mortality (de Moura Gabriel et al. 2022; Passarinho and Barrucho 2021). Even before the direct risks for pregnant women were widely known, women were worried about delivering alone in hospitals and infecting their fetuses and babies amid rising COVID-19 mortality (Meaney et al. 2022).

Rising COVID-19 mortality also led to changes in family dynamics—notably via lockdowns and loss of childcare (Gromada et al. 2020)—that might have also driven down DFS. More broadly, rising mortality corresponded with compounding stress and uncertainty across multiple life domains, which generated anxiety about economic stability, health care, and social support (Sott et al. 2022), which in turn may have contributed to a shifting preference toward smaller families. Indeed, in-depth interviews among young Brazilian women in April–May 2020 revealed how women cited an “enormous amount of fear and worry about COVID-19 and its health and economic consequences as reasons for wanting to delay or avoid pregnancy” (Marteleto and Dondero 2021:2).

An alternate hypothesis is that women may have increased their DFS in response to rising COVID-19 mortality. Prior research has shown how women’s fertility responses are shaped by more than just their individual circumstances and that deaths in one’s immediate vicinity—especially among infants and children—may trigger an insurance motive and compel women to accelerate or increase fertility (Cain 1981; Owoo et al. 2015; Sandberg 2006). Women may also be operating under a community rebuilding framework: when exposed to sudden, large-scale increases in deaths in their communities, women may consider rebuilding and strengthening their community via fertility increases (Nobles et al. 2015). Psychological motivations may also be at play, with women perceiving childbearing as a way of attaining normalcy and meaning in their lives amid sudden large-scale death (Cohan and Cole 2002; Gaspare et al. 2012).

The pathways linking COVID-19 mortality and DFS likely apply differently to women with and without children; to our knowledge, our analysis is the first to underscore this life stage variation for any epidemic. For women with children, increasing COVID-19 mortality may lead to more vigilance of children’s well-being and, consequently, no change or a decrease in DFS. Alternatively, increasing COVID-19 mortality may lead women with children to desire more children than initially reported. With mounting loss, uncertainty, and reminders of the fragility of life, these women may wish to rebuild their communities, secure their legacy, and bring meaning and stability in their lives via childbearing, as seen following other public health shocks (Trinitapoli and Yeatman 2018), natural disasters (Cohan and Cole 2002; Nobles et al. 2015), and violence (Weitzman et al. 2021). Women without children may be similarly inclined to increase DFS when exposed to increasing COVID-19 deaths—in fact, they may have greater flexibility to do so, as they do not have children (Yeatman et al. 2013). However, these women may also revise their DFS downward, as they may be negotiating multiple uncertainties given their earlier life stage (Hayford 2009) and may want to avoid the additional uncertainty of having children.

Data

Our data come from the first two waves of the Demographic Consequences of Zika and Covid Project, an ongoing panel survey following women aged 18–34 in 2020 in Pernambuco, Brazil. The first wave (W1) was fielded in May–October 2020 and the second wave (W2) in May–September 2021. We recruited W1 respondents using random digit dialing with a dual-frame design.1 Per Brazilian census estimates, 94% and 88% of women aged 18–34 in metro Recife and Pernambuco state, respectively, owned a cell phone (IBGE 2021), highlighting the large reach of our sampling base.

In W2, we followed W1 respondents using a hierarchical mixed-method process that included phone calls, web messages and links, WhatsApp messages, and household visits (though interviews could not be conducted face-to-face given COVID-19 protocols). A mixed-mode process increased chances of follow-up with respondents who moved or changed cell numbers. Some 65.8% of W1 respondents (n = 2,624) were reinterviewed in W2. We do not observe selective attrition between waves; W2 respondents do not have statistically different baseline social, demographic, and health characteristics than those not interviewed in W2.2

Surveys were conducted in Portuguese using computer-assisted telephone interviews; each interview lasted 25 minutes, on average. Bilingual staff and the principal investigator translated the questionnaire into English.3 Interviews were recorded for supervision and quality control.

For the analysis, we first pooled W1 and W2 observations for W1 respondents who were reinterviewed in W2. We dropped 104 respondents with missing data on the variables of interest in both waves,4 retaining observations for 2,520 respondents in our final analytic sample. We merged these individual-level data, which include respondent’s municipality of residence in each wave, with municipality-level COVID-19 death rates, calculated using data from Pernambuco.

Methods

We used individual fixed-effects regressions to examine whether change in lagged, municipality-level COVID-19 death rates between waves is associated with change in individual-level DFS. We examined hypothesized variation in responses by motherhood status by including an interaction term between lagged, municipality-level COVID-19 death rates and an indicator for motherhood status in W1. We started with parsimonious models controlling for age, month of interview in W1 and the interval between waves, and survey year. These time controls account for the five-month span within which we conducted interviews in W1, as the pandemic was fast evolving; individual-level variation in duration between W1 and W2 interviews (11 months, on average; standard deviation of two months); and macro-contextual changes across waves. We built on this model with an individual-level control for whether the respondent or her family members had ever had COVID-19, allowing us to uniquely capture how municipality-level changes in COVID-19 mortality influenced DFS net of one’s own or one’s family’s changing COVID-19 infection status.

Our full model includes other time-varying, individual-level controls to account for the multidimensional nature of the pandemic. We controlled for changes in life course factors (marital status, parity) and socioeconomic and health conditions (requested governmental COVID-19 aid between waves, self-rated health).5 This full model further isolates how heightened uncertainty, fear, and community loss linked with rising, local-level COVID-19 mortality informs changing DFS.

Our dependent variable is respondent’s DFS—that is, how many children respondents would choose to have in their life. Following the Demographic and Health Surveys, we asked respondents who had at least one child at the time of interview, “If you could go back to a time when you had no children and could choose the exact number of children you would have in your life, what number would that be?” Respondents who did not have children at the time of interview were asked, “If you could choose the exact number of children you would like to have for your entire life, what number would that be?” For women with children, this measure can be inflated because of post hoc justification of actual family size (McClelland 1983); however, because our analysis focuses on change in DFS between waves, we implicitly account for baseline differences stemming from ex post rationalization among women with children. Further, because our regressions control for children born between W1 and W2, we account for ex post rationalization in women’s DFS from changes in parity between waves.6

Figure 1 shows the proportion of women with and without children at W1 who changed their DFS between waves, as well as average DFS for both groups in W1 and W2. A casual glance at average DFS in W1 versus W2 may suggest no change in family size preferences for women with and without children. However, simply comparing these averages masks the underlying within-person dynamism—that is, the substantial proportion of women who changed their DFS over time. Among women without children in W1, 16% reported higher DFS and 17% reported lower DFS in W2 versus W1. Among women with children in W1, 21% reported higher DFS and 19% reported lower DFS in W2 versus W1.

Fig. 1.

Fig. 1

Open in a new tab

Percentage of respondents who increased, decreased, and did not change their desired family size (DFS) between Wave 1 (W1), 2020, and Wave 2 (W2), 2021

Our main independent variable is the COVID-19 death rate in respondent’s municipality of residence (deaths attributed to COVID-19 per 100,000 people in the municipality) in the month prior to the interview in each wave. We focus on COVID-19 mortality in the month prior as it signifies the extent of COVID-19-related losses and disruptions in the most recent past; these deaths are likely to be salient contextual influences on fertility. The focus on the month prior aligns with the conceptual framing of prior studies that have shown how recent mortality exposure matters for fertility.7

We calculated this using municipality-specific, monthly data on COVID-19 deaths from Brasil.io (2022) that consolidates cause-specific death data from health departments across all states each month. We adjusted one-month lagged COVID-19 deaths for municipality population using the estimated population size for each municipality from a United Nations Development Programme dataset for Brazil (Freire et al. 2019), given Brazil’s lack of census data in 2020. We standardized these rates for regression analysis.

Our measure of municipality-level COVID-19 death rates has sufficient spatiotemporal variation as fieldwork was spread across 94 municipalities and over five months in each wave. Twenty-two percent of women without children in W1 and 17% of women with children in W1 were in a municipality that saw at least a one-standard-deviation increase in COVID-19 mortality between waves.8

Our analysis models within-person change, comparing respondents with themselves in 2020 versus 2021 and, thereby, accounting for time-invariant, individual-level characteristics. Results reflect the association between changing COVID-19 mortality exposure and changing DFS for those with and without children in 2020. They reflect how feelings of stress or worry, uncertainty, and instability that accompany the sudden, increasing, large-scale loss of life in one’s local community are associated with changes in DFS. In Brazil, much of the variation in social distancing measures such as school closures, business closures, and lockdowns were implemented at the state level (Touchton et al. 2021). Our data site is restricted to one state, so we have confidence that our results reflect the implications of rising COVID-19 mortality and associated stressors rather than local-level changes in social distancing measures.

We applied survey weights to adjust for unequal selection probabilities at baseline, to integrate samples, and for nonresponses and coverage adjustments. Standard errors in regression models are clustered at the municipality level.

Results

Table 1 displays the characteristics of our analytic sample respondents. A plurality of respondents were in the older age group in both waves, married in both waves, had children in W1, had good health in both waves, and had themselves or a family member contract COVID-19 in W1. At the same time, respondents reported key changes between waves: 25% had themselves or a family member contract COVID-19, 4% had children, 12% got married, 5% exited a marriage, 14% reported declining health, 10% reported improving health, and 74% requested governmental COVID-19 aid. This highlights the multitude of changes in women’s lives during the pandemic and the importance of an empirical strategy that accounts for such changes to disentangle the effect of COVID-19 mortality exposure on DFS.

Table 1.

Panel respondents’ characteristics in 2020 and 2021, Pernambuco, Brazil, Demographic Consequences of Zika and Covid Project 2020–2021

Characteristic Mean (SD)/%
Average Desired Family Size in W1 1.9 (1.0)
Average Desired Family Size in W2 1.9 (1.0)
Change in Desired Family Size (%)
 No change 63.3
 Increased in W2 18.4
 Decreased in W2 18.3
Average Municipality-Level COVID-19 Death Rate, One-Month Lag, W1 (per 100,000) 22.4 (15.1)
Average Municipality-Level COVID-19 Death Rate, One-Month Lag, W2 (per 100,000) 18.7 (8.5)
Change in Self/Family Member Had COVID-19 (%)
 Never had COVID-19 12.7
 COVID-19 infection in W1 62.5
 COVID-19 infection between W1 and W2 24.8
Change in Age Group (%)
 Less than 25 years in both waves 34.7
 25 years or more in both waves 61.0
 18–24 years in W1 → 25 years or more in W2 4.3
Change in Marital Status (%)
 Not married in both waves 38.7
 Married in both waves 44.1
 Not married in W1 → married in W2 12.0
 Married in W1 → not married in W2 5.2
Had Children in W1 (%) 52.2
Average Number of Children, W1 0.9 (1.1)
Had Child(ren) Between W1 and W2 (%) 3.7
Change in Self-rated Health (%)
 Good in both waves 52.5
 Not good in both waves 23.6
 Good in W1 → not good in W2 14.1
 Not good in W1 → good in W2 9.8
Requested Government Aid in W2 (%) 74.2
Month of Interview in W1 (%)
 May–June 35.8
 July 12.7
 August 20.3
 September–October 31.2
Average Number of Months Between W1 and W2 Interviews 11.1 (1.9)
Unweighted N 2,520
Open in a new tab

Notes: Wave 1 (W1) was fielded in 2020, and Wave 2 (W2) was fielded in 2021. Married individuals include those in formal marriages as well as those in informal unions (living with a romantic partner). We dichotomized our measure of self-rated health for this analysis. “Good” health includes very good and good health; “not good” health includes average, poor, and very poor health. We collected information on government COVID-19 aid only in W2 as the program became more accessible in late 2020. In W1, four respondents were interviewed in May and two respondents were interviewed in October. Panel weights are used.

Table 2 displays results from fixed-effects regressions that examine how change in lagged, municipality-level COVID-19 mortality is associated with change in DFS for women with and without children in W1. Model 1 examines the simple association, using only age and time controls. This model shows that while there is no significant association between COVID-19 mortality and DFS for women without children at W1, there is a significant, positive association for women with children at W1 (significant interaction effect). Rising local COVID-19 mortality is associated with upward revisions in DFS among women with children in W1.

Table 2.

Individual-level fixed-effects regressions examining the association between COVID-19 mortality exposure and women’s desired family size

Model 1 Model 2 Model 3
Lagged Municipality Level COVID 19 Death Rate (standardized; per 100,000) −0.036 −0.039 −0.047
(0.044) (0.041) (0.037)
Lagged Municipality Level COVID 19 Death Rate (standardized; per 100,000) × Had Children in W1 0.133* 0.133* 0.133*
(0.056) (0.057) (0.054)
Age = 25 Years or More −0.205† −0.204† −0.172
(0.122) (0.121) (0.112)
Self/Family Member Had COVID-19 Infection 0.091 0.095
(0.117) (0.106)
Married 0.051
(0.181)
Number of Children −0.338*
(0.155)
Requested Government Aid in W2 0.121
(0.081)
Self-rated Health = Not Good −0.057
(0.074)
Constant 1.658*** 1.709*** 2.104***
(0.308) (0.331) (0.370)
R2 (within) .040 .041 .055
Open in a new tab

Notes: All models use pooled observations from 2,520 respondents. All models include controls for W1 month of interview combined with the interval (months) between the W1 and W2 interviews as well as survey year. All models use panel weights, with standard errors (shown in parentheses) clustered at the municipality level.

Source: Demographic Consequences of Zika and Covid Project surveys, W1 and W2 panel respondents.

†

p < .10;

*

p < .05;

***

p < .001

Model 2 builds on this by controlling for respondent’s own or family’s COVID-19 infection status. Here, we find that the positive relationship between local-level COVID-19 mortality and DFS for women with children at W1 exists even after accounting for individual-level COVID-19 exposure. This highlights how the observed pronatalist responses of women with children were driven not by changes in COVID-19 infection status but by losses experienced in their communities.

Model 3 further explores these associations by including other individual-level controls. As with Models 1 and 2, we find no significant association between the main variables for women without children in W1 and a small but significant positive association for women with children in W1. A one-standard-deviation increase in COVID-19 mortality exposure is associated with a 0.133 net increase in DFS among women with children in W1. Overall, results in Table 2 suggest that there is a different dynamic in DFS at play for women with children: even when we account for co-occurring changes in their conditions (including individual-level exposure to COVID-19), we find that they respond to increasing local-level COVID-19 mortality by adjusting DFS upward.

Results in Table 2 hold across sensitivity and robustness checks. We find similar results when we cap DFS at five or more children, classify COVID-19 death rates into quintiles, and specify our controls differently.9

Discussion

This study shows that young women revised their family size preferences in patterned ways during COVID-19. Thirty-three percent of women without children and 40% of women with children changed their DFS between 2020 and 2021, and these changes reflected meaningful revisions in response to their changing contextual circumstances. Our analysis focused on the influence of changing COVID-19 mortality at the municipality level. We find that women with children responded to increasing COVID-19 mortality by increasing their DFS. This increase holds even after accounting for co-occurring changes in their life course, economic, and health conditions, including their own or their family’s COVID-19 infection status.

There are several possible reasons why women with children increased their DFS in response to growing COVID-19 mortality. When exposed to widespread uncertainty and loss, they may have seen childbearing as a way to rebuild their communities, foster kinship ties, and bring a sense of meaning and stability in their lives; similar responses have been observed following other local mortality shocks, such as natural disasters (Nobles et al. 2015) and terrorist attacks (Rodgers et al. 2005). It is also possible that they were operating under an insurance framework. While COVID-19 disproportionately affected the elderly, Brazil in particular saw rising infant mortality, so women with children may have desired more children to ensure a minimum number of surviving or healthy children (Sandberg 2006). Under either mechanism, it is striking that we find that local COVID-19 mortality had implications for DFS, even beyond women’s own or family’s COVID-19 infection status.

Women without children did not respond similarly to COVID-19 mortality. We speculate that this is because they did not have firsthand experience with childbearing or child-rearing and may have perceived it as a riskier experience during unstable times. When faced with rising deaths and associated uncertainties across multiple life domains, women without children might have preferred to avoid the additional uncertainty of unknown experiences. Another reason might be that women with no children were already aiming for a family size that is low and within a narrow band given Brazil’s low-fertility regime.

Our study makes several important contributions. This is the first analysis to underscore how local-level COVID-19 mortality affected family size preferences. Existing COVID-19 studies have examined how other dimensions of the pandemic—such as economic insecurity (Koenig et al. 2022; Zimmerman et al. 2022), poor health (Manning et al. 2022), and declining relationship quality (Lazzari et al. 2024; Manning et al. 2022)—have influenced fertility preferences. Ours is the first to highlight the significance of COVID-19 mortality for women at different life stages while simultaneously accounting for other changing conditions, including one’s own or family’s infection incidence and life course changes. More broadly, our work speaks to research that conceptualizes fertility preferences as a moving target, sensitive to changing individual and local conditions (Hayford 2009; Trinitapoli and Yeatman 2018; Weitzman et al. 2021).

Our findings build on and align with research documenting a positive relationship between other, non-COVID mortalities and fertility preferences for some women (see studies cited above). To be sure, we recognize that each mortality shock is unique and can have distinct implications for how individuals approach reproduction. For example, the HIV/AIDS epidemic is different from COVID-19 in terms of at-risk age groups and infection transmission modes (Illanes-Álvarez et al. 2021), and these distinctions can translate into different implications for relationship dynamics and fertility preferences. Nonetheless, our findings suggest some broad similarities in how widespread loss, fear of infection or death, and uncertainty during mortality shocks can lead some women to want to (re)build or consolidate their families.

This is also the first study to frame fertility preferences in a dynamic way in a highly unequal, middle-income, Latin American setting. Women in this context were particularly vulnerable to the pandemic, and our findings substantiate how COVID-19 deaths and the associated uncertainties had significant implications for how some women approached one of the most intimate aspects of their lives—childbearing. This fertility response matters because Brazil has witnessed below-replacement fertility over the last decade, with different causes and trends than the shift toward post-modern norms found in Northern Europe.10

Our findings come with some caveats. First, we do not have data on media reporting or knowledge about local COVID-19 mortality and are, therefore, unable to gauge how respondents learned about these deaths. Second, we do not have pre-COVID-19 DFS data for this cohort. Without this benchmark, it is hard to ascertain whether our findings reflect pandemic-specific effects or broader, preexisting fertility trends. Third, while statistically significant and robust, the effect of COVID-19 mortality on DFS is small and any subsequent impact on actual fertility is likely to be limited. This is unsurprising because the effect size reflects average change in the low-fertility context of Brazil (recall that Figure 1 shows a notable share of women revising DFS downward). Fourth, while Pernambuco is fairly similar to the rest of Brazil in terms of fertility levels and COVID-related disruptions to health care services, it fares worse across some socioeconomic indicators (IBGE 2020) and reported some of the country’s highest COVID-19 death rates (Brasil.io 2022). This raises questions about whether we would find similar patterns across other settings within and outside of Brazil. A related question is whether we would find a parallel relationship between COVID-19 deaths and DFS in countries with similarly low fertility but higher ages at first birth. Future work can extend this analysis by considering how mortality in neighboring municipalities informs fertility using spatial modeling.

Overall, this research note offers key insights into how the implications of the pandemic extended beyond health and mortality to fertility. It also provides a framework for considering how mortality shocks in general—and the pandemic in particular—can shape fertility preferences in different ways for women at different life stages. As the pandemic stabilizes and its effects unfold over a longer period, questions arise about whether DFS would return to prepandemic levels or our findings reflect enduring changes. To this end, panel studies tracking fertility preferences, behaviors, and outcomes across different stages of the pandemic and beyond are needed for examining long-term implications for fertility.

Supplementary Material

Supplement

ELECTRONIC SUPPLEMENTARY MATERIAL The online version of this article (https://doi.org/10.1215/00703370-12178940) contains supplementary material.

Acknowledgments

This research was funded by grant R01HD091257, Reproductive Responses to the Zika Virus Epidemic in Brazil, and grant 2R01HD091257-07, Reproductive and Child Health Trajectories in Successive Novel Infectious Disease Crises, both awarded to PI L. J. Marteleto by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD). This research was also supported by grant P2CHD042849, awarded to the Population Research Center at The University of Texas at Austin by the NICHD, as well as grant P2CHD044964, awarded to the Population Studies Center at the University of Pennsylvania by the NICHD. This study was conducted under institutional review board approval #2018-01-0055 from The University of Texas at Austin and is now under institutional review board approval #853752 at the University of Pennsylvania upon the PI’s institutional change, as well as the Brazilian National Commission for Research Ethics (also known as CONEP, or Comissão Nacional de Ética em Pesquisa) study approval CAAE: 34032920.1.0000.5149. Both authors contributed equally to the research note.

Footnotes

1

We used a list-assisted, random digit dialing procedure to implement a dual-frame sample design. For drawing the main sample (70%), we used a sampling base of more than 19 million random cell phone numbers from the Brazilian government’s concession. We used the available 1,000 banks dedicated to cell phones in the target phone area code 81, as informed by the telecommunications authority in Brazil. We then stratified these numbers into three strata. The first two strata were based on region (stratum 1 = metropolitan region of Recife; stratum 2 = nonmetropolitan region of Recife) using the location of the plurality of the listed phones, while the third stratum contained those for whom the 1,000 banks did not have any listed number. The sample was allocated proportionally to the number of 1,000 banks from each stratum. Within strata 1 and 2, the 1,000 banks were selected with probabilities proportionate to the number of listed cell phones; within stratum 3, they were selected at random. All numbers were sampled from the selected 1,000 banks for a total of approximately three million cell phone numbers. We selected the remaining 30% of the sample at random from a commercial database. When drawing our entire sample, we did not use (nor did we have) demographic information (sex or age) of the cell phone owners prior to the call. The numbers were substituted according to our designed protocol, using a system that tracked and classified each call according to the American Association of Public Opinion Research’s (2016) guidelines. Following these definitions, the W1 cooperation rate was 68.9%. More details of the sampling and study procedures are available from the authors.

2

W2 respondents did not differ in terms of baseline age, marital status, parity, DFS, maternal education, race, and self-rated health compared with those not interviewed in W2.

3

The institutional review board at the University of Texas at Austin and the Brazilian National Commission for Research Ethics approved the study. The study is now under the institutional review board at the University of Pennsylvania.

4

Because most of the missing data were for the outcome variable, we would have arrived at a similar final analytic sample size if we were to use multiple imputation methods that impute missing data for independent variables.

5

While self-rated health is a subjective measure, it is widely used in health surveys (including by the World Health Organization) to efficiently measure underlying or actual health conditions. It has consistently proven to be a good predictor of morbidity and mortality (Latham and Peek 2013; Schnittker and Bacak 2014).

6

There is some evidence that DFS—especially prospective measures of DFS—can indicate fertility demand. However, a more complete measure of fertility demand would be based on multidimensional measures that include DFS as well as measures of the certainty and intensity of fertility preferences (Thomson and Brandreth 1995).

7

For example, Weitzman et al. (2023) used local homicide deaths in the week prior to interview, arguing that the cognitive consequences for reproductive outcomes begin to diminish within an expanded window of exposure. Smith-Greenaway et al. (2022:570) used funeral attendance in the month prior, arguing that it serves as a “proxy for the intensity of women’s recent exposure to social network mortality.” Torche and Villarreal (2014) used local homicide deaths six months prior to conception and in specific trimesters of gestation to examine implications for birth outcomes.

8

Thirty percent of women without children and 34% of women with children were in a municipality that saw at least a one-standard-deviation decrease in COVID-19 mortality between waves.

9

This includes defining 27–34 years as the older age group, coding self-rated as a four- or five-category ordinal variable, controlling for depression, and including interactions between W1 motherhood and the other controls.

10

While not this study’s focus, it is worth discussing how our findings show that other shifting conditions influenced DFS. We find that women reduced DFS following births between waves. This may be because women who had a birth in these years faced increased financial and time demands, particularly if caring for a newborn for the first time. These findings provide direction for further work on the many ways the pandemic shaped fertility preferences.

Contributor Information

Letícia J. Marteleto, Department of Sociology and Population Studies Center, University of Pennsylvania, Philadelphia, PA, USA

Sneha Kumar, Human Development and Social Policy Program, Northwestern University, Evanston, IL, USA.

References

  1. Aassve A, Cavalli N, Mencarini L, Plach S, & Sanders S (2021). Early assessment of the relationship between the COVID-19 pandemic and births in high-income countries. Proceedings of the National Academy of Sciences, 118, e2105709118. 10.1073/pnas.2105709118 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Agadjanian V, & Prata N (2002). War, peace, and fertility in Angola. Demography, 39, 215–231. [DOI] [PubMed] [Google Scholar]
  3. American Association for Public Opinion Research. (2016). Standard definitions: Final dispositions of case codes and outcome rates for surveys (Report). Available from https://aapor.org/standards-and-ethics/standard-definitions/
  4. Bankole A, & Westoff CF (1998). The consistency and validity of reproductive attitudes: Evidence from Morocco. Journal of Biosocial Science, 30, 439–455. [DOI] [PubMed] [Google Scholar]
  5. Brasil.io. (2022). COVID-19 epidemiological bulletins by municipality per day [Dataset]. Retrieved from https://brasil.io/dataset/covid19/caso_full/
  6. Cain M (1981). Risk and insurance: Perspectives on fertility and agrarian change in India and Bangladesh. Population and Development Review, 7, 435–474. [Google Scholar]
  7. Celewicz A, Celewicz M, Michalczyk M, Woźniakowska-Gondek P, Krejczy K, Misiek M, & Rzepka R (2023). SARS CoV-2 infection as a risk factor of preeclampsia and pre-term birth: An interplay between viral infection, pregnancy-specific immune shift and endothelial dysfunction may lead to negative pregnancy outcomes. Annals of Medicine, 55, 2197289. 10.1080/07853890.2023.2197289 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Cohan CL, & Cole SW (2002). Life course transitions and natural disaster: Marriage, birth, and divorce following Hurricane Hugo. Journal of Family Psychology, 16, 14–25. [DOI] [PubMed] [Google Scholar]
  9. de Moura Gabriel IW, de Oliveira CGL, Reis AOA, Pereira YTG, Júnior JG, de Matos Brasil AAG,… Neto MLR (2022). Deaths announced in Brazil—A dark sweetness: COVID-19 among children and adolescents. Journal of Pediatric Nursing, 65, e22–e23. 10.1016/j.pedn.2021.12.029 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Diniz D, Brito L, & Rondon G (2022). Maternal mortality and the lack of women-centered care in Brazil during COVID-19: Preliminary findings of a qualitative study. Lancet Regional Health: Americas, 10, 100239. 10.1016/j.lana.2022.100239 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Edlow AG, Castro VM, Shook LL, Kaimal AJ, & Perlis RH (2022). Neurodevelopmental outcomes at 1 year in infants of mothers who tested positive for SARS-CoV-2 during pregnancy. JAMA Network Open, 5, e2215787. 10.1001/jamanetworkopen.2022.15787 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Esteve A, Castro-Martín T, & Castro Torres AF (2022). Families in Latin America: Trends, singularities, and contextual factors. Annual Review of Sociology, 48, 485–505. [Google Scholar]
  13. Freire FHMA, Gonzaga MR, & Queiroz BL (2019). Projeção populacional municipal com estimadores bayesianos, Brasil 2010–2030 [Municipality population projections with Bayesian estimators, Brazil 2010–2030]. In Sawyer DO (Ed.), Seguridade social municipais: Projeto Brasil 3 Tempos [Municipal social security: Brazil 3 Time Periods Project]. Brasília, Brazil: United Nations Development Programme, Brazil and International Policy Centre for Inclusive Growth. [Google Scholar]
  14. Gaspare C, D’Alfonso A, Colagrande I, Catana P, Casacchia M, & Patacchiola F (2012). Post-earthquake birth-rate evaluation using the brief cope. Journal of Maternal-Fetal and Neonatal Medicine, 25, 2411–2414. [DOI] [PubMed] [Google Scholar]
  15. Gonçalves BMM, Franco RPV, & Rodrigues AS (2021). Maternal mortality associated with COVID-19 in Brazil in 2020 and 2021: Comparison with non-pregnant women and men. PLoS One, 16, e0261492. 10.1371/journal.pone.0261492 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gromada A, Richardson D, & Rees G (2020). Childcare in a global crisis: The impact of COVID-19 on work and family life (Innocenti Research Brief, No. 2020–18). UNICEF Office of Research–Innocenti. Retrieved from https://www.unicef.org/innocenti/media/6121/file/UNICEF-IRB-Childcare-in-a-Global-Crisis-2020.pdf [Google Scholar]
  17. Hayford SR (2009). The evolution of fertility expectations over the life course. Demography, 46, 765–783. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Heiland F, Prskawetz A, & Sanderson WC (2008). Are individuals’ desired family sizes stable? Evidence from West German panel data. European Journal of Population / Revue Européenne de Démographie, 24, 129–156. [Google Scholar]
  19. Heuveline P, & Poch B (2007). The phoenix population: Demographic crisis and rebound in Cambodia. Demography, 44, 405–426. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Iacovou M, & Tavares LP (2011). Yearning, learning, and conceding: Reasons men and women change their childbearing intentions. Population and Development Review, 37, 89–123. [DOI] [PubMed] [Google Scholar]
  21. Illanes-Álvarez F, Márquez-Ruiz D, Márquez-Coello M, Cuesta-Sancho S, & Girón-González JA (2021). Similarities and differences between HIV and SARS-CoV-2. International Journal of Medical Sciences, 18, 846–851. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Instituto Brasileiro de Geografía e Estatística. (2020). Pesquisa Nacional por Amostra de Domicílios [National Household Sample Survey; ] [Dataset]. Available from https://www.ibge.gov.br/estatisticas/sociais/trabalho/9171-pesquisa-nacional-por-amostra-de-domicilios-continua-mensal.html [Google Scholar]
  23. Instituto Brasileiro de Geografía e Estatística. (2021). Pesquisa Nacional por Amostra de Domicílios Contínua—PNAD [Continuous National Household Sample Survey—PNAD; ] [Dataset]. Available from https://www.ibge.gov.br/estatisticas/sociais/educacao/17270-pnad-continua.html?edicao=27138&t=o-que-e [Google Scholar]
  24. Instituto Brasileiro de Geografía e Estatística. (2024). Population Projections for Brazil and Federation Units: 2000–2070 [Dataset]. Retrieved from https://www.ibge.gov.br//en/statistics/social/population/18176-population-projection.html
  25. Koenig LR, Smith Hughes C, Gemmill A, & Diamond-Smith N (2022). Unequal impacts of the COVID-19 pandemic on women’s fertility preferences: Evidence from a longitudinal cohort in the United States. SSM–Population Health, 20, 101305. 10.1016/j.ssmph.2022.101305 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Latham K, & Peek CW (2013). Self-rated health and morbidity onset among late midlife U.S. adults. Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 68, 107–116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Lazzari E, Reimondos A, & Gray E (2024). Did the COVID-19 pandemic affect fertility desires in Australia? Understanding why people changed their attitudes towards having a first or additional child. Population and Development Review, 50(S1), 243–276. [Google Scholar]
  28. Lindberg LD, VandeVusse A, Mueller J, & Kirstein M (2020). Early impacts of the COVID-19 pandemic: Findings from the 2020 Guttmacher Survey of Reproductive Health Experiences (Report). Guttmacher. Retrieved from https://www.guttmacher.org/sites/default/files/report_pdf/early-impacts-covid-19-pandemic-findings-2020-guttmacher-survey-reproductive-health.pdf [Google Scholar]
  29. Luppi F, Arpino B, & Rosina A (2020). The impact of COVID-19 on fertility plans in Italy, Germany, France, Spain, and the United Kingdom. Demographic Research, 43, 1399–1412. 10.4054/DemRes.2020.43.47 [DOI] [Google Scholar]
  30. Manning WD, Guzzo KB, Longmore MA, & Giordano PC (2022). Cognitive schemas and fertility motivations in the United States during the COVID-19 Pandemic. Vienna Yearbook of Population Research, 20, 261–284. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Marteleto LJ, & Dondero M (2021). Navigating women’s reproductive health and childbearing during public health crises: COVID-19 and Zika in Brazil. World Development, 139, 105305. 10.1016/j.worlddev.2020.105305 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Marteleto LJ, Guedes G, Coutinho RZ, & Weitzman A (2020). Live births and fertility amid the Zika epidemic in Brazil. Demography, 57, 843–872. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. McClelland GH (1983). Family size desires as measures of demand. In Bulatao RA & Lee RD (Eds.), Determinants of fertility in developing countries (Vol. 1, pp. 288–343). New York, NY: Academic Press. [Google Scholar]
  34. Meaney S, Leitao S, Olander EK, Pope J, & Matvienko-Sikar K (2022). The impact of COVID-19 on pregnant womens’ experiences and perceptions of antenatal maternity care, social support, and stress-reduction strategies. Women and Birth, 35, 307–316. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Miller WB, Barber JS, & Gatny HH (2013). The effects of ambivalent fertility desires on pregnancy risk in young women in the USA. Population Studies, 67, 25–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Müller MW, Hamory J, Johnson-Hanks J, & Miguel E (2022). The illusion of stable fertility preferences. Population Studies, 76, 169–189. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Nobles J, Frankenberg E, & Thomas D (2015). The effects of mortality on fertility: Population dynamics after a natural disaster. Demography, 52, 15–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Owoo NS, Agyei-Mensah S, & Onuoha E (2015). The effect of neighbourhood mortality shocks on fertility preferences: A spatial econometric approach. European Journal of Health Economics, 16, 629–645. [DOI] [PubMed] [Google Scholar]
  39. Passarinho N, & Barrucho L (2021, April 14). Why are so many babies dying of COVID-19 in Brazil? BBC News. Retrieved from https://www.bbc.com/news/world-latin-america-56696907 [Google Scholar]
  40. Rocca CH, Doherty I, Padian NS, Hubbard AE, & Minnis AM (2010). Pregnancy intentions and teenage pregnancy among Latinas: A mediation analysis. Perspectives on Sexual and Reproductive Health, 42, 186–196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Rocca CH, Parra M, Muñoz I, Foster DG, Boscardin WJ, & Ralph LJ (2022). Comparison of pregnancy preferences preceding vs year 1 of the COVID-19 pandemic. JAMA Network Open, 5, e2220093. 10.1001/jamanetworkopen.2022.20093 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Rodgers JL, St. John CA, & Coleman R (2005). Did fertility go up after the Oklahoma City bombing? An analysis of births in metropolitan counties in Oklahoma, 1990–1999. Demography, 42, 675–692. [DOI] [PubMed] [Google Scholar]
  43. Sandberg J (2006). Infant mortality, social networks, and subsequent fertility. American Sociological Review, 71, 288–309. [Google Scholar]
  44. Schnittker J, & Bacak V (2014). The increasing predictive validity of self-rated health. PLoS One, 9, e84933. 10.1371/journal.pone.0084933 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Sennott C, & Yeatman S (2012). Stability and change in fertility preferences among young women in Malawi. International Perspectives on Sexual and Reproductive Health, 38, 34–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Smith-Greenaway E, Yeatman S, & Chilungo A (2022). Life after loss: A prospective analysis of mortality exposure and unintended fertility. Demography, 59, 563–585. 10.1215/00703370-9807961 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Sobotka T, Jasilioniene A, Galarza AA, Zeman K, Németh L, & Jdanov D (2021). Baby bust in the wake of the COVID-19 pandemic? First results from the new STFF data series (Working paper). Retrieved from https://osf.io/mvy62_v1/download/
  48. Sobotka T, Skirbekk V, & Philipov D (2011). Economic recession and fertility in the developed world. Population and Development Review, 37, 267–306. [DOI] [PubMed] [Google Scholar]
  49. Sott MK, Bender MS, & da Silva Baum K (2022). COVID-19 outbreak in Brazil: Health, social, political, and economic implications. International Journal of Health Services, 52, 442–454. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Thomson E, & Brandreth Y (1995). Measuring fertility demand. Demography, 32, 81–96. [PubMed] [Google Scholar]
  51. Torche F, & Villarreal A (2014). Prenatal exposure to violence and birth weight in Mexico: Selectivity, exposure, and behavioral responses. American Sociological Review, 79, 966–992. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Touchton M, Knaul FM, Arreola-Ornelas H, Porteny T, Sánchez M, Méndez O,… Rosado JA (2021). A partisan pandemic: State government public health policies to combat COVID-19 in Brazil. BMJ Global Health, 6, e005223. 10.1136/bmjgh-2021-005223 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Trinitapoli J, & Yeatman S (2011). Uncertainty and fertility in a generalized AIDS epidemic. American Sociological Review, 76, 935–954. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Trinitapoli J, & Yeatman S (2018). The flexibility of fertility preferences in a context of uncertainty. Population and Development Review, 44, 87–116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Weitzman A, Barber J, Heinze J, Kusunoki Y, & Zimmerman M (2023). Exposure to nearby homicides and young women’s reproductive lives during the transition to adulthood. American Journal of Sociology, 129, 856–906. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Weitzman A, Barber JS, Heinze J, & Zimmerman M (2021). How nearby homicides affect young women’s pregnancy desires: Evidence from a quasi-experiment. Demography, 58, 927–950. 10.1215/00703370-9160045 [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Yeatman S (2009). HIV infection and fertility preferences in rural Malawi. Studies in Family Planning, 40, 261–276. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Yeatman S, Sennott C, & Culpepper S (2013). Young women’s dynamic family size preferences in the context of transitioning fertility. Demography, 50, 1715–1737. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Zimmerman LA, Karp C, Komuro N, Akilimali P, Zakirai MS, OlaOlorun F,… Gummerson E (2023). Change in fertility intentions in the first year of COVID-19: Evidence from four countries in sub-Saharan Africa. Population and Development Review, 50(S1), 177–211. [Google Scholar]
  60. Zimmerman LA, Karp C, Thiongo M, Gichangi P, Guiella G, Gemmill A,… Bell SO (2022). Stability and change in fertility intentions in response to the COVID-19 pandemic in Kenya. PLoS Global Public Health, 2, e0000147. 10.1371/journal.pgph.0000147 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement
NIHMS2149790-supplement-Supplement.pdf (69KB, pdf)

RESOURCES