Human Reproduction, Vol.24, No.6 pp. 1492– 1500, 2009
Advanced Access publication on February 26, 2009 doi:10.1093/humrep/dep040
ORIGINAL ARTICLE Reproductive epidemiology
Trends and seasonality in birth
frequency: a comparison of Muslim and
Jewish populations in southern Israel:
daily time series analysis of 200 009
births, 1988 –2005
Michael Friger 1, Ilana Shoham-Vardi, and Kathleen Abu-Saad
Department of Epidemiology and Health Services Evaluation, Faculty of Health Sciences, Ben-Gurion University of the Negev, Beersheba,
Israel
1
Correspondence address. E-mail: [email protected]
background: The aim of this study was to assess effects of natural and artificial seasonality on reproduction in a large cohort of Muslim
and Jewish women, living in the same geographic area.
methods: Population included all births between January 1988 and December 2005 (6613 daily records of 200 009 births) at the Soroka
University Medical Center (southern Israel). A ‘classical’ time series technique based on generalized linear regression models was used.
results: A systematic increase in the number of births during Ramadan was observed in the Muslim (P , 0.001), but not in the Jewish,
population. This ‘Ramadan effect’, adjusted for trend and seasonal factors, was statistically significant only in Muslim multiparae (P , 0.001),
where the model explained more than 48% of the variance, compared with 7% in Jewish multiparae. Seasonal birth variations in Muslim
primiparae were not associated with Ramadan and were similar to those in Jewish primiparae. There were no differences in length of gestation associated with Ramadan.
conclusions: The findings suggest an increased conception rate following the Hajj pilgrimage, which support the hypothesis that seasonality of reproductive behavior is influenced by socio-cultural factors more than by geographic and climatic factors, and can help plan preconception interventions in Muslim populations in southern Israel.
Key words: birth seasonality / environmental effects / epidemiology / mathematical modeling / pregnancy
Introduction
Time-of-year cyclic changes have been shown to affect human reproduction (Smits et al., 1998; Bobak and Gjonca, 2001; Cesario, 2002;
Mikulecky and Lisboa, 2002; Cancho-Candela et al., 2007). These
cyclic changes include both natural cycles which are defined by geophysical factors, and climatic characteristics, and artificial cycles which are
determined by socio-cultural, behavioral and economic factors, such as
weekly cycles, calendar-month cycles, vacation-working time cycles
and holiday cycles.
Significant seasonal and annual cyclic variation in the number of births
has been described for different populations and geographic zones (Macfarlane, 1974; Guptill et al., 1990; Lam and Miron, 1991; Lam et al., 1994;
Scott et al., 1998), and biological (Boklage et al., 1992; Lam and Miron,
1994; Merrow et al., 2005), physiological, physical, chemical and
psychological mechanisms have been suggested as explanations
(Cesario, 2002; Cancho-Candela et al., 2007). As Macfarlane (1974) has
shown, seasonality in conception rates is more pronounced in populations
living in regions with dramatic changes in climate than in geographic regions
where climate remains more or less stable, such as subarctic and equatorial geographic zones. The various effects of artificial seasonality were
studied in Brazil (Mikulecky and Lisboa, 2002), Switzerland (Lerchl and
Reinhard, 2008) and Israel (Cohen, 1983) with regards to day of the
week, and in the USA with regards to vacation and working time cycles
(Rosenberg, 1966). Holidays and festivals have been shown to impact
number of births, which peak on Christmas (Rosenberg, 1966; Wrigley
and Schofield, 1981) and on Yom Kippur in Israel (Cohen, 1983; Omer
et al., 1986; Wiser et al., 1997).
Unlike Muslim holidays, Christian and Jewish holidays coincide with
natural seasons. Christian holidays are determined by the solar calendar,
& The Author 2009. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved.
For Permissions, please email: [email protected]
1493
Birth frequency trends in Muslim and Jewish populations
whereas Jewish holidays are determined by the lunar calendar, which is
corrected every 4 years to correspond to the natural seasonality.
Muslim holidays are determined only by the lunar calendar, thus move
forward 11 days each year in relation to the solar calendar, and therefore
are not associated with natural seasonality (Riazuddin and Khan, 2005).
The aim of the present study was to study the effect of natural and artificial seasonality on reproduction in a large cohort of Muslim women, with
special emphasis on the month of Ramadan. During the month of
Ramadan, Muslims are required to fast from dawn to sunset (www.fastramadan.com/Ramadan_brief_introduction.php, accessed October 2008).
Pregnant, lactating or menstruating women are not required to fast but
have to make up for it by fasting at a later time, but many pregnant
women choose to fast with their families rather than having to fulfill the
fasting duty on their own after they give birth (Joosoph et al., 2004).
Based on the assumption that most pregnant women fast during
Ramadan, we hypothesized that births would peak annually during the
month of Ramadan in Muslim but not in Jewish women, due to shortened
gestation, which may be caused by possible dehydration (Stan et al.,
2002). A unique opportunity for this study occurs in southern Israel
(Negev), where the birth population includes both Muslim and Jewish
women all of whom give birth in Soroka University Medical Center
(SUMC). Comparing reproductive annual cycles in Jewish and Muslim
populations living in the same area allows for establishing the net effect
of the month of Ramadan, not confounded by natural seasonality.
Materials and Methods
The database used for the current work was derived from a computerized
file of individual birth records of SUMC that serves the entire population
of southern Israel. The derived database consists of sequential time series
of daily records including: total number of births, number of births to primiparae (first births) and multiparae (births 1), number of preterm births
(,258 days of gestation) and term births (259 days of gestation), and
number of births to Jewish and to Muslim women. In addition, each day
was classified as either a Ramadan day or a non-Ramadan day. These
records have been collected over 18 years between January 1988 and
December 2005 (in total: 6613 daily records, representing 200 009 births).
Due to our interest in identifying periods with potential importance for the
timing of periconceptional public health interventions to improve birth outcomes, the data set included all births (e.g. stillbirths, multiple births, etc.).
Statistical analysis
For analysis of the considered time series, we used a ‘classical’ time series
additive model: decomposition of the time series in the sum of the three
components: trend, periodicity and ‘white noise’. Generally, the trend
represents a long-term tendency, the periodicity component represents
regular oscillations about a trend and the ‘white noise’ is a random component, an error of the modeling. We describe trend by using cubic polynomial smoothing. We model the trend of time series Yt by polynomial
regression and the trend Tr(Yt) is presented in the following polynomial
form: Tr(Yt) ¼ a0 þ a1. t þ a2. t 2 þ a3. t 3, where t is a time variable.
For the presentation of the periodic component of the time series, we
use the pairs of sine and cosine functions representing different cycles in a
P
form known as a Fourier series: ki¼1 ðbi cosðvi tÞ þ ci sinðvi tÞÞ, where k is a
number of considered cycles, t is a time variable (in days), each pair of
cosine– sine functions represents one cycle, and vi denotes frequency of
the periodic oscillation with period Ti days: vi ¼ 2p/Ti (i ¼ 1, 2, . . ., k).
Our strategy is to test the significance of the following a priori hypothesized
cycles: year, half-year, season, months, lunar month and week. Let
us denote the annual (year) period by T1 (365.25 days), then T2 ¼ T1/2
is the half-year cycle period, T3 ¼ T1/4 is a 3-month seasonal
period (four seasons per year), T4 ¼ T1/3 is a 4-month seasonal
period (three seasons per year), T5 ¼ T1/12 is the 1-month period, T6is the weekly cycle period (7 days) and T7 is a lunar-month cycle
period (28 days). (This regression model is sometimes called a
Cosinor model.) The time series is represented in the following form:
P
Yt ¼ a0 þ a1 t þ a2 t2 þ a3 t3 þ 7i¼1 ðbi cosðvi tÞ þ ci sinðvi tÞÞ þ 1t ; where
e t is a ‘white noise’. We will call this model a ’trend-harmonic’ model
(TH model). A similar technique was used in van Straten et al., (2008).
We performed this time series approach using Generalized Linear
Model regressions from the Gaussian or/and Poisson families. Gaussian
models were used for fitting ‘count’ outcome variables when the sample
was large and the means of the variables were large (.9), such as daily
number of births and daily number of births to multiparae.
As previous research has shown different distribution of births in primiparae
and multiparae (Guptill et al., 1990), we fit daily time series to births of primiparae and multiparae in the Jewish and Muslim populations, separately.
The final models included only significant trend and seasonal patterns;
however, we included the dichotomous Ramadan variable in all models
as a variable of special interest. Results were considered to be of statistical
significance if the relevant P-value was 0.05. Statistical analyses were performed using the Statistical Package for the Social Sciences (version 14)
and STATA (version 9.2).
Results
The overall description of the daily time series of numbers of births in
general, and births to primiparae and to multiparae women, over the
period of 1988 –2005 in the Jewish and Muslim populations is given in
Table I.
The lowest daily average number of births was observed in 1988 for
both populations. In the Jewish population, the daily average number
Table I Daily numbers of births by parity in the Jewish and Muslim populations in southern Israel (1988 –2005)
Jewish population
.........................................................................
Total
Primiparae
Multiparae
Muslim population
.........................................................................
Total
Primiparae
Multiparae
.............................................................................................................................................................................................
Mean
SD
Median
15.2
3.6
4.5
2.0
15.0
3.0
11.6
15.1
2.4
12.7
3.8
5.7
1.7
5.1
11.0
15.0
2.0
12.0
Minimum
3
0
1
1
0
1
Maximum
33
13
28
39
12
34
1494
of births was 13.4 (median 13), 2.9 (median 3) and 10.5 (median 10)
for all births, births to primiparae and births to multiparae, respectively; and in the Muslim population, the numbers were 9 (median
9), 1.5 (median 1) and 7.6 (median 7) for all births, births to primiparae and births to multiparae, respectively. The highest daily
average number of births in the Jewish population was in 2003: 16.7
Friger et al.
(median 17), 4.3 (median 4) and 12.4 (median 12) for all births,
births to primiparae and births to multiparae, respectively. For the
Muslim population, the mean of daily total number of births and the
daily mean of births to primiparae reached its peak in 2003: 20.4
(median 20) and 3.1 (median 3), respectively. The highest mean of
daily births to Muslim multiparae was in 2003: 17.7 (median 19).
Figure 1 Daily mean number of births by year and parity in Muslim and Jewish populations (1988– 2005) in southern Israel.
Figure 2 Daily mean number of births by Gregorian calendar months and parity in Muslim and Jewish populations in southern Israel (1988– 2005).
Birth frequency trends in Muslim and Jewish populations
The dynamics of the birth number per year by population and parity
group is presented in Fig. 1. Figure 2 presents daily mean number of
births by Gregorian calendar months and parity in the Muslim and
Jewish populations, and Fig. 3 presents the same data but on a yearly basis.
1495
Examination of the mean birth distributions by Gregorian calendar
months for Muslims and Jews reveals a marked difference in the temporal behavior of the Jewish and Muslim populations, attributable
mainly to the difference between the two groups of multiparae.
Figure 3 Observed and predicted daily mean number of births by years and parity in the Muslim and Jewish populations in southern Israel (1988–2005).
1496
Friger et al.
Muslim multiparae births tend to peak in the winter months, whereas
in the Jewish population, the number of multiparae births in the last 6
months of the year tends to be higher than that in the first 6 months of
the year (Fig. 2).
Results of the analysis of trends and a priori hypothesized periodicities of the total number of births for both populations (Jewish and
Muslim, Bedouin) are presented in Table II.
Unlike the Jewish population, where a positive linear trend is
evident, the overall growth tendency in number of births in the
Muslim population is larger and ‘sharper’, and the linear trend component is absent, resulting in a non-linear growth (Fig. 1 and
Table II). Some of the significant periodic patterns (annual, half-year
and weekly cycles) are the same for both populations, but they
differ in the slopes (amplitudes) and phases of the cycles as well as
in the periodic behavior of the time series (Table II). While the
3-month seasonal cycle is significant in the Muslim population, it is
non-significant in the Jewish population. The overall amplitude of the
daily time series in the total number of births in the Muslim population
is higher than in the Jewish population (Fig. 2).
It is important to note that our TH model explains more than 49%
of variance in the Muslim population but only 10% in the Jewish population (Table II). Thus, the model of the daily time series of number of
births expresses seasonal behavior for both populations, but it is more
evident and better explains the trend in the Muslim population than in
the Jewish population (Fig. 2).
The results of the TH analysis show that there is a systematic
increase in the number of births during Ramadan in the Muslim population (Fig. 4) but not in the Jewish population (Table II). This
‘Ramadan effect’ in the Muslim population is significant, adjusted for
trend and all considered seasonal factors.
Table III shows results of the TH analysis for the daily time series of
primiparae in both the Jewish and Muslim populations. A Poisson
regression model was used because the daily means of births of primiparae were small. In both populations of primiparae, the time series
had the same periodic patterns; the trends were weak, non-linear
and almost stationary. The results of the analysis (Table III), and the
graphic presentation (Figs 3 and 4) of primiparae time series show
that the ‘Ramadan effect’ (Table III), adjusted for trend and all considered seasonal factors, in both populations is non-significant and
the predictive ability of the TH-model is low (Table III).
Table IV shows the results of the TH analysis for the daily time
series in both groups of multiparae. Unlike the results of the time
series analysis in primiparae, in multiparae there is a distinct difference
between Jews and Muslims. For the Muslim multiparae population, the
time series has an additional significant periodic pattern: 4-month seasonal cycle, which is not significant either for Jewish multiparae or for
all births (primiparae and multiparae together) in both populations
(Table II). Moreover, the ‘Ramadan effect’ is significant for the
Muslim and non-significant for the Jewish multiparae population. The
results of the analysis are shown in Table IV and presented graphically
in Fig. 4. These TH models explain more than 48% of variance in the
Muslim population but only 7% in the Jewish population. The multiparae daily time series of births clearly show seasonal behavior (halfyear and weekly) for both populations, but the patterns are more
explicit, more pronounced and better explained in the Muslim
population.
Table II Results of trend-harmonic analysis for daily total number of births
Variables
Muslim population (adjusted R 2 5 0.49)
Jewish population (adjusted R 2 5 0.10)
Regression coefficients
Regression coefficients
.............................................................
Sig (P)
.............................................................
Sig (P)
.............................................................................................................................................................................................
Trend
Factor significance
Yes
Factor significance
Yes
intercept
9.243
,0.001
13.146
,0.001
0.001
,0.001
day
0.001
day2
7.76E2007
,0.001
0.338
25.25E2008
0.587
day3
28.83E2011
,0.001
28.74E2012
0.364
Annual cycle
sin(vy.day)
Factor significance
Yes
Factor significance
Yes
21.277
,.001
20.858
,0.001
cos(vy.day)
1.294
,.001
20.031
Half-year cycle
cos(vhy.day)
Factor significance
Yes
Factor significance
Yes
0.494
,0.001
0.293
,0.001
0.678
sin(vhy.day)
0.276
,0.001
0.293
,0.001
Seasonal cycle
cos(vs3.day)
Factor significance
Yes
Factor significance
No
Weekly cycle
cos(vw.day)
Factor significance
Yes
0.327
sin(vw.day)
0.293
Ramadan effect
Ramadan (1—yes, 0—no)
0.198
0.005
0.113
0.128
Factor significance
No
,0.001
21.400
,0.001
,0.001
0.092
Factor significance
Yes
Factor significance
0.916
,0.001
0.229
vy is an annual cycle frequency, vhy is a half-year cycle frequency, vs3 is a 3-month season cycle frequency and vw is a weekly cycle frequency.
0.214
No
0.244
1497
Birth frequency trends in Muslim and Jewish populations
significant differences between Ramadan and non-Ramadan births in
the proportions of stillbirths (1.0% and. 1.0%, respectively), multiple
births (3.7% and 3.1%, respectively) and Cesarean sections (11.3%
and 11.7%, respectively).
Discussion
Figure 4 Daily mean number of births within and out of Ramadan
period by parity in the Muslim population in southern Israel (1988–
2005).
To explore the possibility that the increase in number of births
during Ramadan days is caused by shortened gestation, we compared
the average length of gestation of births during the Ramadan to that of
the rest of the year in the Muslim population. No clinically significant
differences were observed in the population examined: the average
length of gestation of births during Ramadan was 273.67 + 16.92,
and during the rest of the year was 273.32 + 16.55. There were no
Our study, which was conducted on two large population-based birth
cohorts employing techniques of time series analysis, shows significant
systematic differences in cycles and trends of the daily number of
births between the Muslim and Jewish populations living in the same
geographic area and giving birth at the same medical center. The
main contribution to these differences is the reproductive pattern of
Muslim multiparae, and specifically, a systematic increase in daily
number of births during the month of Ramadan. Our data also
suggest that the effect of Ramadan has become more pronounced
over time. These effects are clearly verified by fitted models, after
adjusting for other seasonal parameters and trends. To the best of
our knowledge, these findings were not previously reported.
The annual increase in number of births during Ramadan is consistent with our initial hypothesis, but the absence of evidence of shortened length of gestation in births during the Ramadan does not
support our initial hypothesized explanation that the excess number
of birth is caused by fasting and dehydration (Stan et al., 2002). No
observed excess of preterm births or shortened gestation during
Ramadan was observed. Moreover, the excess of births during
Ramadan was only observed in Muslim multiparae, not in primiparae.
The lack of negative physiological effects among pregnant women
fasting during Ramadan was reported by several other studies conducted in different populations. A large cohort study compared the
birthweights of 13 351 full-term infants of Asian Muslim mothers
with the birthweights of infants born to white and non-Muslim Asian
babies in Birmingham during the month of Ramadan, and found no
differences between Muslim and non-Muslim infants (Cross et al.,
1990). In several smaller cross-sectional and case –control studies
that examined the nutritional status of fasting and non-fasting expectant mothers, no effects of fasting were found either in the volume
of amniotic fluid (Kiziltan et al., 2005) or uterine artery flow (Mirghani
et al., 2007). The only significant physiologic effect reported to be
associated with fasting during Ramadan was a lower rate of fetal
breathing movements (Mirghani et al., 2004).
An alternative explanation of the observed increase in the number
of births during Ramadan in our study may be the result of an increase
in the number of conceptions occurring 9 months prior to Ramadan.
Examination of the Muslim calendar reveals that this period coincides
with the time the Hajj, which occurs during the first 10 days of the
12th month of the Islamic year. The Hajj (pilgrimage to the holy city
of Mecca), which all Muslims aspire to perform at least once in their
lifetime, if they can afford it financially, is one of the five religious obligations of observant Muslims. During the Hajj, pilgrims have to keep
themselves pure, and thus must totally abstain from sexual activity
(http://www.al-bab.com/arab/background/hajj.htm). The excess
numbers of births seen in Ramadan, we suggest, is the result of
increased number of conceptions occurring right after the Hajj. We
could not obtain information about the annual number of pilgrims
from southern Israel going on Hajj. As there has been a significant temporal trend of intensification of religiosity in the Muslim population in
1498
Friger et al.
Table III Results of trend-harmonic analysis for daily number of births to primiparae*
Variables
Muslim population
.......................................................
Rate Ratio
Sig
Jewish population
.......................................................
Rate Ratio
Sig
.............................................................................................................................................................................................
Trend
Factor significance
Yes
Factor significance
Yes
intercept
1.48
,0.001
2.93
,0.001
day
1.00(þ)**
0.067
1.00(þ)
0.271
day2
1.00(þ)
0.029
1.00(þ)
0.085
3
day
1.00(2)***
Annual cycle
sin(vy.day)
Factor significance
Yes
Factor significance
Yes
0.85
,0.001
0.93
,0.001
cos(vy.day)
0.80
,0.001
0.98
0.089
Half-year cycle
cos(vhy.day)
Factor significance
Yes
Factor significance
1.06
,0.001
1.02
Factor significance
Yes
Factor significance
Seasonal cycle
cos(vs3.day)
0.001
1.02
Weekly cycle
cos(vw.day)
0.98
Factor significance
Ramadan effect
Factor significance
Ramadan (1—yes, 0—no)
0.98
0.033
Yes
0.032
No
0.462
1.00(2)
0.005
Yes
0.035
Yes
1.02
Factor significance
0.031
Yes
0.94
,0.001
Factor significance
No
1.00
0.937
vy is an annual cycle frequency, vhy is a half-year cycle frequency, vs3 is a 3-month season cycle frequency and vw is a weekly cycle frequency.
*Time series of primiparae births fitted by the Poisson regression.
**Value close to 1 but .1.
***Value close to 1 but ,1.
Table IV Results of trend-harmonic analysis for daily number of births to multiparae
Variables
Muslim population (adjusted R 2 5 0.48)
Jewish population (adjusted R 2 5 0.07)
Regression coefficients
Regression coefficients
.............................................................
Sig
.............................................................
Sig
.............................................................................................................................................................................................
Trend
Factor significance
Yes
Factor significance
intercept
7.680
,0.001
10.188
,0.001
day
0.0001
0.001
,0.001
day2
6.46E2007
,0.001
21.50E2007
0.075
day3
27.30E2011 (21.165)
,0.001
5.40E2012
0.518
Annual cycle
sin(vy.day)
cos(vy.day)
Factor significance
Yes
Factor significance
Yes
20.914
,0.001
20.588
,0.001
1.848
,0.001
0.027
Half-year cycle
cos(vhy.day)
Factor significance
Yes
Factor significance
Yes
0.348
,0.001
0.228
,0.001
sin(vhy.day)
0.208
0.001
0.285
,0.001
Seasonal cycle
cos(vs3.day)
Factor significance
Factor significance
No
0.138
0.032
0.042
0.516
sin(vs4.day)
0.129
0.046
20.012
0.856
Weekly cycle
cos(vw.day)
Factor significance
Yes
0.027
sin(vw.day)
0.228
Ramadan effect
Ramadan (1—yes, 0—no)
0.313
Yes
Yes
0.679
Factor significance
Yes
,0.001
21.165
,0.001
,0.001
0.084
Factor significance
Yes
Factor significance
0.945
,0.001
0.203
0.190
No
0.235
vy is an annual cycle frequency, vhy is a half-year cycle frequency, vs3 is a 3-month season cycle frequency, vs4 is a 4-month season cycle frequency and vw is a weekly cycle frequency.
1499
Birth frequency trends in Muslim and Jewish populations
Israel (Aburaiya, 2004), we assume that there has been an increase in
the number of persons going on Hajj. This is consistent with the finding
that the Ramadan effect becomes evident since about 1998.
This explanation is also consistent with the observation that there is a
marked difference in temporal behavior between the first and subsequent births in the Muslim population. An earlier study based on
the birth distribution of Negev Bedouins in 1981–1982 (Guptill et al.,
1990) reported different patterns of birth seasonality of primiparae
and multiparae. Our data for the period of 1988–2005 clearly show
a consistent difference between Muslim primiparae’s and multiparae’s
temporal reproductive behavior. In the Jewish population, which
served as a comparison population, there is no such marked difference
between primiparae and multiparae in terms of birth seasonality.
The Bedouin-Muslim population places a very high value on fertility;
large families are considered a blessing, as well as a source of pride,
and a large number of children are also a determinant of power and
social status. The Bedouin-Muslim population in southern Israel has
a higher fertility rate than any other population group in Israel. The
total fertility rate in the Bedouin-Muslim population in 2006 was 7.3,
when compared with other Muslim populations in Israel (3.3 –4.1)
and with the Jewish population in southern Israel (2.62). While
Bedouin-Muslims constitute about one-fifth of the Negev population,
they contribute about half of the births. Family planning and contraceptive use were found to be relatively uncommon during the time
series period we studied (Hundt et al., 2000). Women are married
at a relatively young age and expected to bear children soon after marriage, and to continue childbearing throughout their fertile years. Thus,
the timing of the first birth is determined largely by the time of marriage. A seasonality of marriage has been reported for this population,
with most marriages taking place in the late spring and summer
(Guptill et al., 1990), and we actually observed an increased number
of primiparae births 12 months later. There are no Islamic religious
edicts determining the seasonality of marriage; however, among the
Negev Bedouin it is customary not to marry during the period
between the feast at the end of Ramadan (Eid al-Fitr) and the feast
celebrated at the culmination of the Hajj (Eid El Adha). This period
(covering 2 months and 10 days) does not cover the entirety of the
marriage season (3 and a half to 4 months), and the cycle that
brings it into the marriage season only reoccurs every 24 years.
Thus, it is not likely to consistently affect marriage seasonality, and
we did not find any evidence of such an effect upon the periodicity
of primiparae births. The timing of conception in multiparae is influenced by several factors, including natural seasonality and seasonality
determined by religious and cultural factors.
The strengths of this population-based study are as follows: a large
sample size, the ability to compare two populations living in the same
area and, hence, subject to the same natural, but different sociocultural artificial, seasonality. As the study population includes practically all births in a defined geographical area, there is no selection
bias. The main limitation of the study is the lack of individual information about particular religious practices (fasting in Ramadan, or
going on the Hajj).
In conclusion, the results of our TH analysis show that two populations living in the same geographic (and climatic) zone have different
seasonal reproductive behavior. Seasonal patterns of births to multiparae in the Muslim population are much more pronounced and
more explicit (the TH model explains more than 48% of the variance)
than the seasonal patterns of the time series of births to multiparae in
the Jewish population (the TH model explains 7% of the variance).
These findings support the hypothesis that the seasonality of reproductive behavior is influenced more by socio-cultural factors than by
geographic (and climatic) factors.
As a growing emphasis in recent years has been placed upon the life
course approach to women’s health, and inter-pregnancy and preconception care (Lu, 2008), this information can help planners of interpregnancy, prenatal and post-natal services to focus and time their
interventions. For example, campaigns to promote the perinatal use
of folic acid to reduce neural tube defects (CDC, 2008) can be
timed prior to Eid El Adha, the Muslim holiday which is celebrated
at the culmination of the Hajj.
References
Aburaiya I. The 1996 split of the Islamic Movement in Israel: between the
holy text and the Israeli-Palestinian context. International Journal of
Politics, Culture and Society 2004;17:439 – 455.
Bobak M, Gjonca A. The seasonality of live birth is strongly influenced by
socio-demographic factors. Hum Reprod 2001;16:1512 – 1517.
Boklage CE, Kirby CF, Zincone LH. Annual and sub-annual rhythms in
human conception rates. I. Effective correction and use of public
record LMP dates. Int J Fertil 1992;37:74– 81.
Cancho-Candela R, Andres-de Llano J, Ardura-Fermandez J. Decline and
loss of birth seasonality in Spain: Analysis of 33,421,731 births over
60 years. J Epidemiol Community Health 2007;61:713 – 718.
Center for Disease Control and Prevention (CDC). Use of supplements
containing folic acid among women of childbearing age – United States.
MMWR 2008;11:5 – 8.
Cesario SK. The ‘Christmas Effect’ and other biometeorologic influences
on childbearing and the health of women. J Obstet Gynecol Neonatal
Nurs 2002;31:526 – 535.
Cohen A. Seasonal daily effect on the number of births in Israel. J R Stat Soc
Ser C Appl Stat 1983;32:228 – 235.
Cross JH, Eminson J, Wharton BA. Ramadan and birth weight at full term
in Asian Muslim pregnant women in Birmingham. Arch Dis Child 1990;
65:1053– 1056.
Guptill K, Berendes H, Forman MR, Chang D, Sarov B, Naggan L,
Hundt GL. Seasonality of births among Bedouin Arabs residing in the
Negev Desert of Israel. J Biosoc Sci 1990;22:213– 223.
Hundt GL, Beckerleg S, Kassem F, Abu Jafar AM, Belmaker I, Abu-Saad K,
Shoham-Vardi I. Women’s health custom made: building on the 40 days
postpartum for Arab women. Health Care Women Int 2000;21:529– 542.
Joosoph J, Abu J, Yu SL. A survey of fasting during pregnancy. Singapore
Med J 2004;45:583 – 586.
Kiziltan G, Karabudak E, Tuncay G, Avsar F, Tuncay P, Mungan O, Meral P.
Dietary intake and nutritional status of Turkish pregnant women during
Ramadan. Saudi Med J 2005;26:1782 – 1787.
Lam DA, Miron JA. Seasonality of birth in human populations. Soc Biol
1991;38:51 – 78.
Lam DA, Miron JA. Global patterns of seasonal variation in human fertility.
Ann N Y Acad Sci 1994;709:9 – 28.
Lam DA, Miron JA, Riley A. Modeling seasonality in fecundability,
conceptions, and births. Demography 1994;31:321 – 346.
Lerchl A, Reinhard SC. Where are the Sunday babies? II. Declining
weekend birth rates in Switzerland. Naturwissenschaften 2008;
95:161– 164.
Lu MC. We can do better: improving women’s health in America. Curr
Opin Obstet Gynecol 2008;20:563– 565.
1500
Macfarlane WV. Seasonal cycles of human conception. Prog Biometeorol
1974;1:557 – 577. 711 – 713.
Merrow M, Spoelstra K, Roenneberg T. The circadian cycle: daily rhythms
from behavior to genes. EMBO Rep 2005;6:930– 935.
Mikulecky M, Lisboa H. Daily birth number in Passo Fundo, South Brazil,
1997–1999: trends and periodicities. Braz J Med Biol Res 2002;35:985–990.
Mirghani HM, Weerasinghe SD, Smith JR, Ezimokhai M. The effect of
intermittent maternal fasting on human fetal breathing movements.
J Obstet Gynaecol 2004;24:635– 637.
Mirghani HM, Salem M, Weerasinghe SD. Effect of maternal fasting on
uterine arterial blood flow. J Obstet Gynaecol Res 2007;33:151 – 154.
Omer D, Friedlander D, Palti Z, Shekel I. Life stresses and premature
labor: Real connection or artifactual findings? Psychosom Med 1986;
48:362 – 369.
Riazuddin R, Khan MUH. Detection and forecasting of Islamic calendar
effects in time series data. State Bank of Pakistan (SBP) Research
Bulletin 2005;1:24 – 34.
Rosenberg H. Seasonal Variation of Births, United States, 1933 – 63. Vital
Health Stat 21, 1 – 59. Washington, DC: National Center for Health
Statistics, 1966.
Friger et al.
Scott S, Duncan SR, Duncan CJ. The interacting effects of prices and
weather on population cycles in preindustrial community. J Biosoc Sci
1998;30:15 – 32.
Smits L, Zielhuis G, Jongbloet P, Straatman H. Seasonal variation in human
fecundability. Hum Reprod 1998;13:3520 – 3524.
Stan G, Boulvain C, Hirsbrunner-Amagbaly M, Pfister P. Hydration for
treatment of preterm labour. Cochrane Database Syst Rev 2002;
CD003096.
Wiser A, Maymon E, Shoham-Vardi I, Silberstein T, Wiznitzer A, Katz M.
Effect of the Yom Kippur fast on parturition. Harefuah 1997;
132:745– 748 (Heb) 824 (English abstract).
Wrigley E, Schofield R. The Population History of England, 1541 – 1871: A
Reconstitution. Cambridge, MA: Harvard University Press, 1981.
van Straten M, Shpigel NY, Friger M. Analysis of daily body weight of
high-producing dairy cows in the first one hundred twenty days of
lactation and associations with ovarian inactivity. J Dairy Sci 2008;
91:3353 – 3362.
Submitted on December 23, 2008; resubmitted on January 25, 2009; accepted on
January 29, 2009