Seasonality of pregnancies, births and marriages in Malta

Introduction

Seasonality of births has been noted in various countries, including in Malta.^1–3 Indeed, an earlier study in Malta had shown that for the period 1950-1996 there was a significant peak in marriages paralleled by a peak in births which lagged after the peak in marriages by 13-14 months. The peak in births was caused by first pregnancies only.³

This study was carried out to ascertain whether seasonality at birth has persisted in relation to first pregnancies, i.e., first time mother has been pregnant and delivered at 22+ weeks of pregnancy, never having suffered a previous miscarriage). Furthermore, potential deviations from expected norms due to the COVID-19 pandemic were also sought.^4,5

Methods

Data

Anonymous aggregated data for births by month and birth order for 1999-2020 was obtained from the National Obstetric Information System, Directorate for Health Information and Research. Monthly marriages were obtained from Malta’s National Statistics Office. Since this is an anonymous and aggregated dataset, ethical approval was not needed, and neither was informed consent, for the same reason.

Analysis

The following time series were studied: First pregnancies (as defined above) and Total marriages. Analyses were performed with unprocessed monthly time series and seasonally adjusted time series.

Seasonal adjustment refers to a technique wherein independent trends in time series data are preserved after removing seasonal variations. It employs a series of mathematical and statistical operations, such outlier detection, model identification, decomposition etc., which can be specified in the software.

Seasonality in this study was analysed using JDemetra+ software (version 2.2.2) with Tramo-Seats and RSA5 specifications. JDemetra+ provides seven predefined specifications for both TRAMO/SEATS and X-13 and uses some default specifications for each method.

The data's seasonality was analysed by the TRAMO/SEAT and RSA5 method. It is a model-based seasonal adjustment method that, in the first step, detects deterministic calendar variation and outliers, and corrects for missing values whilst applying a regression model to the data. T trading day effect and the leap year effect are pre-tested and taken in account. A Malta-specific calendar was used, with pre-specified national holidays.

The regression ARIMA (linear, autoregressive integrated moving average) model chosen was multiplicative with parameters of (3,1,1)(0,1,1) for Marriages and (0,0,0)(0,1,1) for First pregnancies.

In further steps, the following seasonality tests were carried out: the Friedman test for stable seasonality, the Kruskal-Wallis test, for seasonality presence assuming stability, and evaluative seasonal test for detecting the presence of identifiable seasonality. These tests were merged within the final combined seasonality test, which determines whether seasonality is present within the series.

A direct seasonal adjustment approach was carried out with the monthly series First pregnancies, from January 1999 to December 2020. The data series Total marriages was seasonally adjusted for January 1999 to February 2020, due to end-of-series outliers between March-December 2020 caused by COVID-19 effects.

A p value ≤0.05 was taken to represent a statistically significant result.

Results

Seasonally adjusted and unadjusted data for the variables in studied are shown in Table 1. Both series exhibited minor seasonal variation, as per Figure 1 and Figure 2. The time series for Marriages exhibits a peak of seasonal variation in June and in September (Table 2). The First pregnancies (Table 3) series exhibits peaks of seasonal variation in September and March.

Figure 1

Seasonal patterns for Malta 1999-2020: Marriages

Figure 2

Seasonal patterns for Malta 1999-2020: First Pregnancies

The actual, non-seasonally adjusted data for Marriages and First pregnancies were compared, with forecasted values for March-December 2020, as can be seen in Figure 3 and Figure 4. The time series 1999-February 2020 served as a base for the forecast. From August 2020, there was a constant, negative gap between the actual data and forecasted values for the three series. Figure 3 illustrates the difference.

Figure 3

Observed data for Marriages 1999-2020 and forecasted values for March-December 2020

Figure 4

Observed data for First Pregnancies 1999-2020 and forecasted values for March-December 2020

Marriages in Malta dropped 72% during March-August 2020, compared to the expected forecasted values. Similarly, First pregnancies dropped by 8% during September-December 2020. In December, the decrease reached 11% for First pregnancies.

Discussion

The drop in marriages and first pregnancies was temporally associated with the brunt of the COVID-19 pandemic, implying a negative impact on both.^4,5

Seasonal patterns in births occur almost universally due to cultural and/or biometeorological factors.

Examples of cultural factors have included festivals, such as a September birth spike probably due to an increase in coital activity in the Christmas period.⁶ Biometeorological factors include agricultural cycles which have been implicated due to potential swings in maternal nutritional status.⁷ Fluctuations may also be due to changes in hours of sunlight and/or temperature.⁸ The best known seasonal patterns occur in the southern United States, with a birth decline in April and May, and in northern Europe with a birth peak in March/April.^1,9

A previous paper based on Maltese data had shown a late summer peak in first births only, lagging behind marriages by approximately 10 months,³ an effect that was similarly noted in Japan.¹⁰ Indeed, seasonal birth patterns have been noted in many countries including birth peaks in Portugal in May and September,¹¹ and July–September peaks in Croatia,¹² The absence of such seasonality is rare and notable.¹³ The Maltese peak had been linked with marriages,³ a link that was also described in other countries.^14–16

The Maltese peak may potentially be attributed to conscious family planning with abstention or contraception prior to marriage, and this may be due to a residual Roman-Catholic ethos.¹⁷ Alternatively, this may be due to simple economics with couples preferring to start families once they have formally settled down contractually in wedlock.¹⁷ Furthermore, marriages in Malta are often large-scale, and are considered important family and social occasions. Naturally, an outdoor venue is generally preferred in clement Mediterranean weather.¹⁸ Thus, a socio-cultural bias to Spring weddings influences first pregnancies and births in Malta.

Marriages in Malta correlate with first pregnancies, resulting in seasonal patterns for both. The COVID-19 pandemic has reduced marriages and first pregnancies, implying that the global events which appeared in March 2020 also affected the decisions of having first pregnancies.

References

1. Lam DA, Miron JA. Seasonality of births in human populations. Soc Biol. 1991;38(1–2):51–78.
2. Grech V, Gatt M. Assisted reproductive technology and multiple pregnancy in Malta: a population-based study. Early Hum Dev. 2021;157:105378.
3. Grech V, Savona-Ventura C, Agius-Muscat H, Janulova L. Seasonality of births is associated with seasonality of marriages in Malta. J Biosoc Sci. 2002;35:(1)95–105.
4. Velavan TP, Meyer CG. The COVID-19 epidemic. Trop Med Int Health. 2020;25:278–80.
5. Cuschieri S, Calleja N, Devleesschauwer B, Wyper GMA. Estimating the direct COVID-19 disability-adjusted life years impact on the Malta population for the first full year. BMC Public Health. 2021;21.(1)
6. Cowgill UM. Season of birth in man: contemporary situation with special reference to Europe and the Southern Hemisphere. Ecology. 1966;47:(4)614–23.
7. Mosher SW. Birth seasonality among peasant cultivators: the interrelationship of workload, diet, and fertility. Hum Ecol. 1979;7:(2)151–81.
8. Roenneberg T, Aschoff J. Annual rhythm of human reproduction: I. Biology, sociology, or both? J Biol Rhythms. 1990;5:(3)195–216.
9. Russell D, Douglas AS, Allan TM. Changing seasonality of birth: a possible environmental effect. J Epidemiol Community Health. 1993;47:(5)362–7.
10. Matsuda S, Kahyo H. Geographical differences and time trends in the seasonality of birth in Japan. Int J Epidemiol. 1994;23:(1)107–18.
11. Caleiro A. Exploring the peaks and valleys in the number of births in Portugal. Hum Ecol. 2010;38:(1)137–45.
12. Polašek O, Kolčić I, Vorko-Jović A, Kern J, Rudan I. Seasonality of births in Croatia. Coll Antropol. 2005;29:(1)249–55.
13. Pascual J, García-Moro C, Hernández M. Non-seasonality of births in Tierra del Fuego (Chile). Ann Hum Biol. 2000;27:(5)517–24.
14. Stoeckel J, Choudhury AK. Seasonal variation in births in rural East Pakistan. J Biosoc Sci. 1972;4:(1)107–16.
15. Malina RM, Himes JH. Seasonality of births in a rural Zapotec municipio, 1945–1970. Hum Biol. 1977;49:(2)125–37.
16. Holland B. Seasonality of births: stability and change in a developing country. Hum Biol. 1989;61:(4)591–8.
17. Savona-Ventura C. Family planning in a Roman Catholic community. Plan Parent Eur. 1995;24:(1)20–2.
18. Chetcuti D. The climate of the Maltese Islands: a review. Msida: University of Malta; 1992.

Figure

Test image