One hundred years ago in 1919: New Zealand’s birth reduction shock associated with an influenza pandemic

The influenza pandemic that caused an estimated 9,000 deaths in late 1918 in New Zealand is generally well documented.1–3 While a fertility impact on the Māori population was suggested by Pool in 1973 (with an estimate of 12.2% of marriages in the 25–34 year age-group dissolved due to death of spouses4), the impact on natality in New Zealand from this pandemic has never been considered in any substantive way. We therefore aimed to conduct a preliminary analysis of this likely association given the 100-year historical point, its potential future relevance to pandemic planning, and its potential relevance to modern-day recommendations for pregnant women to be vaccinated against seasonal influenza.

Methods

Official data covering the period 1910 to 1930 were sourced from multiple New Zealand Yearbooks,5 including those published up to 1932. But due to limitations with the denominator data for the Māori population around this time period, we use modelled Statistics New Zealand estimates for calculating rates (specifically from Figure 2.2 in a Report6).

To put our findings into a broader demographic context we considered data on marriages and post-First World War troop movements back to New Zealand. As only annual marriage data were available, to give an idea of monthly marriage trends we ran queries of marriages from the national ‘Births, Deaths and Marriages’ database.7 These queries used the common names (applied to the names of both brides and grooms) of: ‘Smith’, ‘Wilson’ and ‘Brown’. A full copy of all the data in an Excel file is available on request.

Results

The number of births declined in both 1918 and 1919 for the non-Māori (essentially European) and Māori populations (Figures 1 and 2 show numbers, since we considered rates to be less reliable due to concerns with denominator accuracy for the Māori population). In 1919 there were 3,756 fewer non-Māori and 239 fewer Māori births than the pre-pandemic year of 1917. Similarly, the reductions in birth rates per 1,000 population in 1919 were by 16.6% and 19.8% respectively, relative to 1917. The birth rate reductions in 1918, relative to 1917, were less at 8.8% and 6.7% reductions respectively. In 1920 the birth rate had returned to near the 1917 level for non-Māori (ie, only 2.3% lower than in 1917), but declined further for Māori (21.9% lower, although it started to rise again in 1921; Figure 2 for numbers).

Figure 1: Numbers of non-Māori births in New Zealand per year from 1910 to 1930.

‍Figure 2: Numbers of Māori births in New Zealand per year from 1910 to 1930 (albeit with missing data for the first four years as registration of Māori births only became a legal requirement in March 1913 and probably rose initially due to improvements in the registration process).

Possible impacts of deaths of women and men in relevant age-ranges

The detailed analysis by Rice1 reports that 2,217 non-Māori women aged 15–44 years died from the pandemic in 1918. Yearbook data from the 1916 census indicated that 53.0% of this age-group were married and had an annual fertility rate of 189.5 per 1,000 population. As a result of these sudden pandemic-related deaths among women in the fertile age-range an estimated 223 children would not have been born per year (ie, 2,217x53.0%x189.5/1,000). Furthermore, this demographic effect is equivalent to 5.9% of the missing births for non-Māori in 1919 (ie, 223/3,756), albeit a simplistic analysis that does not account for pregnant women probably being at increased risk of pandemic-associated death (see elsewhere8 for evidence on this association).

Similar calculations are not so readily reproducible for married men given the unusual distribution of New Zealand men related to the First World War effort. That is, in November 1918, an estimated 22% of New Zealand men aged 20–49 years were still overseas (Table 1). Also other men were in military training camps in New Zealand, and in predominantly male populated mining towns as part of producing coal and scheelite for the war effort. Given such complexities, we simplistically assumed the same value as for women above at 5.9% of missing births being potentially due to the deaths of men who would otherwise have fathered children in 1919. Despite how New Zealand men were distributed geographically as per above, this estimate may be conservative since 1.9 times more non-Māori men aged 15–44 died in the pandemic than non-Māori women, with 1,372 extra male deaths.1

Table 1: Numbers of New Zealand military personnel overseas in 1918 and 1919* in relation to the First World War.

*Data from the 1919 Yearbook for numbers in 1918 and for the return rate that allowed the estimates for 1919 ie, based on the statement that: “An average of 4,500 men per month from the UK and France, and 500 men per month from Egypt, were returned to New Zealand from the date of the Armistice.”
**The percentages use the New Zealand male population aged 20–49 years from the 1916 Census.

Possible impact on sexual behaviour and subsequent conceptions

At this period in New Zealand’s social development, marriage was fairly closely aligned with natality. That is in 1919 only 4.6% of births (1,132/24,483) were deemed “illegitimate” according to the 1920 Yearbook. Also, although there were substantive New Zealand troop movements (Table 1), the substantial post-war upturn in marriage did not clearly start until mid-1919 (Figure 3). At an annual level the number of non-Māori marriages in 1917 was 6,417 and this declined slightly in the pandemic year of 1918 (to n=6,227, a 3.0% decline), but then rose again in 1919 (n=9,519, a 33% increase on 1917 numbers) and surged further in 1920 (to n=12,175). The increase in marriages in 1919 was clearly not enough to counter the natality decline in this particular year. But reduced sexual activity among couples at the time the pandemic struck is a plausible contributor to some reduced natality in the subsequent year of 1919. This effect is likely because the peak pandemic months of November and December in 1918 caused substantive social disruption in New Zealand (eg, from illness affecting much of the population and people travelling to care for sick relatives and to attend funerals). This disruption in turn might logically have resulted in a decline in coital frequency for sexually active couples in these two peak pandemic months, along with spouses potentially avoiding close contact to prevent infecting each other. If such an overall decline in coital frequency was at a 50% level for these two peak pandemic months, then it might be assumed that this would average out to around 8% less sexual activity during the whole 12-month time period: end of March 1918 to the start of April 1919 (ie, (100%-[12-(2x50%)/12])=8.3%). This is the time period when conceptions would have occurred which subsequently generated births within the 1919 calendar year, all else being equal.

Figure 3: Numbers of marriages by month for 1917 to 1919 using a sample of three common surnames in the national ‘Births, Deaths and Marriages’ database given the absence of national-level data (see Methods).

Possible impact of embryonic and fetal loss

To estimate the potential embryonic and fetal loss contribution (for loss in all three trimesters) to the reduced birth rate in 1919 relative to the 1917 year for non-Māori, we subtracted the two categories detailed above for parental death and reduced sexual activity. This gave the estimate of 79.9% of the missing births in 1919 relative to the 1917 year (ie, [100%-((5.9%x2)+8.3%)]=79.9%). This decline is equivalent to around 3,000 fewer births associated with embryonic and fetal loss among non-Māori in 1919 (3,756x79.9%=3,001).

Discussion

It is well established that influenza infection is associated with embryonic/fetal loss and that influenza vaccination in pregnancy reduces rates of stillbirths, eg, based on a systematic review.9 Therefore, it seems very likely that the birth rate reductions we have described above for New Zealand in 1918 and 1919 reflect such losses, albeit with some smaller roles for adult deaths and reduced sexual activity. Such birth rate reductions associated with the 1918 pandemic have also been observed in other settings. For example, there was a 5–15% decline for the US and Scandinavia at 6–7 months post-pandemic10 and a 43% reduction at 9–11 months after the peak of pandemic mortality for one county in Arizona, US.8 Another US study11 estimated a 10% drop 9–10 months after peak influenza mortality, which the authors ascribed to a reduction in conception during the period of intense pandemic activity. This was in addition to a birth rate reduction in the three months after peak mortality, which they associated with excess preterm births and stillbirths from influenza infections in the last trimester. Other work has also demonstrated post-pandemic birth rate reductions in Taiwan12 and Sri Lanka.13

Our estimates for the natality impacts for 1919 may actually be underestimates of the extent of the loss when using 1917 as the comparison year for at least two reasons. Firstly men returning from war during late 1918 and 1919 brought with them an estimated 3,000 wives back to New Zealand from England according to the 1919 Yearbook (some of whom would have been pregnant and boosted the official 1919 birth rate in New Zealand). Secondly, the adult male population in New Zealand in late 1918 and 1919 would have risen above the 1917 level as men returned from the war zone (Table 1). Even though the marriage rate did not seem to pick up until mid-1919 (Figure 3), some of these returning men would already have been married prior to going to war—and so a resumption of sexual activity and pregnancy of their partners would be expected soon after their return to New Zealand (and so potentially contributing to births in late 1919).

The larger natality shock for Māori in 1919 compared with non-Māori is consistent with Māori experiencing a disproportionately higher mortality burden from this pandemic (and in two subsequent pandemics14). This difference again highlights the importance of current efforts to eliminate poverty and ethnic inequalities in health in New Zealand, as well as more specific strategies to protect Māori health (eg, via eliminating tobacco, controlling the obesogenic environment and improving access to healthcare).

After the initial post-pandemic natality decline for Māori, the data also suggests further sudden dips in 1923 and 1926 (Figure 2). We have no definitive explanation for these subsequent dips, except to note that declines in births have been reported for subsequent waves of the 1918 pandemic elsewhere (ie, in 1920 in the US11).

It should be noted that our analysis is still fairly preliminary and if more research was undertaken then epidemiological modelling and time-series analyses could be performed (ideally using more fine-grained data that could arise from collating individual birth registrations or tracking cohorts of couples and returning soldiers). This analysis could then more accurately estimate the relative roles of embryonic/fetal loss along with uncertainty distributions. Such knowledge could then better inform planning around future influenza pandemics. This response includes the potential need to prioritise protection of pregnant women from infection via: protective sequestration, use of any scarce pandemic vaccines or antivirals, or prioritised access to ventilators in hospital ICUs. Such knowledge could also inform the design of current day informational materials to promote the routine use of influenza vaccination by pregnant women against seasonal influenza in New Zealand. The World Health Organization gives a high priority to vaccinating pregnant women and New Zealand has fully funded influenza vaccination for pregnant women since 2010. While the current rate of influenza vaccination in pregnancy in New Zealand is not accurately established, it is thought to be low, eg, one unpublished analysis suggested it was under 30% in 2016 (Howe A, New Zealand Influenza Symposium, February 2019, Wellington). Mechanisms to improve this low current coverage need to be explored, along with efforts to maximise uptake by population groups with the highest needs: particularly Māori, Pasifika and low-income New Zealanders.