COVID-19 vaccination was linked to almost 70,000 fewer US live births in 2023

CDC data from 566 US counties, with a combined population of nearly 260 million, show that COVID-19 vaccination in 2023 was significantly linked to fewer live births. Extrapolating these findings to the entire US population estimates 69,598 fewer births (95% CI: -111,215; -27,981), representing a 1.90 percent reduction (95% CI: -3.01; -.785). Additionally, the vaccine was linked to fewer births in 2022, but the association was not statistically significant that year.

In a previous post, I showed that COVID-19 vaccination caused over 138,000 US deaths in 2022 and over 150,000 in 2023. Hence, COVID-19 vaccination has decimated the US population by increasing deaths and decreasing births, which follow-up research should explain.

A concern beyond what I described above is that the vaccine seems to have induced an increasing trend in deaths and a decreasing trend in births, which follow-up research should monitor and eventually also explain.

Descriptive statistics

Table 1 reports US live birth rates (live births per capita) from 566 counties for 2020, 2021, 2022, and 2023, relative to 2018-2019 (counties’ live births in 2018 plus 2019 divided by the population size in 2018 plus 2019), multiplied by 100 (Note 1). It shows fewer births after 2018-2019, particularly in 2023. The increased standard deviation, particularly in 2023, implies greater variation in births across counties.

Table 1. Descriptive statistics for counties’ birth rates relative to 2018-2019, multiplied by 100, weighted by 2022 population size. The table presents data from 566 large-population counties.

Regressions

Table 2 reports regressions where the dependent variables are counties’ 2022 (Model 1) and 2023 (Model 2) live birth rates relative to 2018-2019. The independent variables are per-capita COVID-19 vaccination uptake by the end of 2021 (Model 1) and 2022 (Model 2), also multiplied by 100 (Note 2).

Table 2. Regressions with robust standard errors, weighted by counties’ population sizes. The dependent variables are the 2022 (Model 1) and 2023 (Model 2) live birth rates relative to 2018-2019, multiplied by 100.

Conservative two-tailed tests of significance concerning the regression coefficients. † p < .10; * p < .05; ** p < .01; *** p< .001. 95% CIs in parentheses. Variance inflation factors (VIFs) in brackets concerning per-capita vaccine uptake by the end of 2021 (Model 1) and 2022 (Model 2), respectively.

The counties from which the CDC reports birth statistics are included, minus three that did not include vaccine data by the end of 2021 (Model 1) and four by the end of 2022 (Model 2). The models control for lagged dependent variables (Note 3).

Model 1 shows a negative, but non-significant, association between vaccine uptake by the end of 2021 and 2022 live births relative to 2018-2019. The 2021 (2020) live birth rates relative to the 2018-2019 control variable are significantly (non-significantly) associated with the dependent variable.

Model 2 shows a strongly significant negative association between vaccine uptake by the end of 2022 and 2023 live births relative to 2018-2019. I.e., the higher the counties’ vaccine uptake, the fewer the live births. The 2022 and 2021 live birth rates, relative to the 2018-2019 control variables, are significantly associated with the dependent variable, but the effects decreased over time. The 2020 live birth rates relative to the 2018-2019 control variable are not significantly associated with the dependent variable.

Very low variance inflation factors (VIFs) concerning the independent variables show that multicollinearity is not a problem.

Margins effects

The CDC reported COVID-19 vaccination data from 3,144 US counties by the end of 2022, covering almost the entire US. The weighted average uptake from those counties by the end of 2022 was 194.3 doses per capita, and using that number as an input in Stata’s post-estimation margins effects command (based on Model 2, Table 2), returned a value of 93.0 (CIs are reported in Table 3). I.e., the county-level weighted-average US COVID-19 vaccine uptake by the end of 2022 predicts a birth rate of 93.0 relative to 2018-2019 (when controlling for lagged dependent variables). Assuming zero vaccine uptake, on the other hand, predicts a birth rate of 94.8. I.e., the birth rate is 1.90 percent lower under a zero-vaccine assumption than under average vaccine uptake. As the US reported 3,596,017 live births in 2023, the reduction in births was estimated to be 69,598 (calculations were carried out using Stata’s nlcom algebra function on the margins post-estimation).

Table 3. Margins post-estimations (based on Model 2, Table 2) comparing birth rates to zero and population-weighted average per-capita Covid-19 vaccination uptake in 3,114 US counties with a population of 329,462,485.

Notes

1. The CDC reports birth data from 570 large-population counties only. For consistency, I omitted four of them for all years that did not include vaccine data by the end of 2021 and 2022 (cf. Note 2).

2. Vaccine data were included from counties reporting positive values on Completeness_pct. To model a proxy for doses per capita, I first summarized the number of doses administered in each county for Administered_Dose1_Recip, Series_Complete_Yes (which typically includes two doses), Booster_Doses, Second_Booster_50Plus, and Bivalent_Booster_5Plus. Next, I divided the number by the population sizes in 2021 and 2022, respectively, and multiplied the result by 100. I.e., per capita vaccine uptake refers to the number of doses administered per 100 people over a given period. One county reporting more than 500 doses administered per 100 by the end of 2022 was omitted from the 2023 analysis.

3. The motive is that the approach “provides a simple way to account for historical factors that cause current differences in the dependent variable that are difficult to account for in other ways”, according to Wooldrige, probably among the most authoritative voices in econometrics today. Including more than one lagged dependent variable, if available, is useful. Issues causing differences in 2022 and 2023 live birth rates relative to 2018-2019, of which we are unaware, may have also contributed to similar differences in previous years. For example, counties experiencing an abnormally low (high) number of births in 2018 and 2019 have led to estimating relatively high (low) rates in the following years. Including birth-rates relative to 2018-2019 in the years preceding 2022 or 2023 as control variables, this approach controls for such issues. Another way to explain the inclusion of lagged dependent variables is that it keeps the birth-rates relative to 2018-2019 in years before the estimation year fixed. That is, a model estimates the vaccine effect on birth rates relative to 2018-2019, for instance, 2022, while canceling out differences between counties in 2020 and 2021. It helps isolate the vaccine effect we are interested in. Not doing that would instead increase noise in the data, as counties with high or low rates relative to 2018-2019 in 2020 and 2021 may experience the same in 2022. That would lead to overestimations (underestimations) of counties with high (low) 2022 values. For a further discussion of the pros of including the lagged dependent variable and cons in a few instances, see, for example, Rönkkö.