Accepted Manuscript

Determinants of different birth intervals of ever married women: Evidence from

Bangladesh

Benojir Ahammed, Rasel Kabir, Menhazul Abedin, Mohammad Ali, Akhtarul Islam

PII: S2213-3984(18)30029-0
DOI: https://doi.org/10.1016/j.cegh.2019.01.011
Reference: CEGH 330

To appear in: Clinical Epidemiology and Global Health

Received Date: 28 January 2018

Revised Date: 3 November 2018
Accepted Date: 30 January 2019

Please cite this article as: Ahammed B, Kabir R, Abedin M, Ali M, Islam A, Determinants of different birth
intervals of ever married women: Evidence from Bangladesh, Clinical Epidemiology and Global Health
(2019), doi: https://doi.org/10.1016/j.cegh.2019.01.011.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Title Page

Determinants of Different Birth Intervals of Ever Married Women: Evidence

from Bangladesh

PT
RI
Benojir Ahammed, M.Sc.1*, Md. Rasel Kabir, MS2, Md. Menhazul Abedin, M.Sc.3, Mohammad
Ali, M.Sc.4 and Md. Akhtarul Islam, MS5

SC
1
Statistics Discipline, Khulna University, Khulna-9208, Bangladesh, Email: [email protected]
2
Statistics Discipline, Khulna University, Khulna-9208, Bangladesh, Email: [email protected]
3
Statistics Discipline, Khulna University, Khulna-9208, Bangladesh, Email: [email protected]
4
Statistics Discipline, Khulna University, Khulna-9208, Bangladesh, Email: [email protected]

U
5
Statistics Discipline, Khulna University, Khulna-9208, Bangladesh, Email: [email protected]
AN
This research did not receive any specific grant from funding agencies in the public, commercial,
or not-for-profit sectors.
M

* Corresponding author:
Benojir Ahammed*;
D

Statistics Discipline,
Khulna University, Khulna-9208, Bangladesh.
TE

Email: [email protected]
Mobile: +8801714960969
C EP
AC
 ACCEPTED MANUSCRIPT

Determinants of Different Birth Intervals of Ever Married Women:

Evidence from Bangladesh
Abstract
Introduction: The analysis of factors affecting birth interval helps to reveal the mechanism and
dynamics of fertility behavior which changes the population structure of a country.

PT
This study aims to examine the association among different socioeconomic and demographic
factors with birth intervals and to identify the statistically significant factors of birth intervals.

RI
Material and Methods: This study has been used latest Bangladesh Demographic and Health
Survey 2014 data based on two stage stratified sample of households. Different birth intervals

SC
have been analyzed using various statistical techniques such as Kaplan Meier estimator, log rank
test, Cox proportional hazard model and Frailty model.
Results: Among the ever married women, 92.40%, 76.80% and 61.60% have experienced live

U
birth for first, second and third times, and the estimated median duration of birth intervals are
AN
26, 46, and 58 months, respectively. Using the log rank test, this study found that region, place of
residence, mothers education, partners education, contraceptive use, and mothers age are
significantly associated with birth intervals. This study also considered mothers age and working
M

status at first birth interval; mothers education, age, working status and survival status of
previous child at second birth interval; mothers education, sex of household head, working
D

status, mass media exposure, age, contraceptive use, religion, and survival status of previous
TE

child at third birth interval are protective factors in next births.

Conclusions: Overall, mothers age, working status and mass media exposure are protective
factors in birth intervals. To fulfill the target, government should pay attention to significant
EP

protective factors.
C

Key words: Birth interval, censored observation, ever married women, survival analysis.
AC

Page 1 of 16
 ACCEPTED MANUSCRIPT

1. Introduction
Nowadays over population is the most alarming problem. Excess population growth is a big and
challenging problem in Bangladesh and this growth varies significantly due to social, regional
and educational factors. One of the important determinants of population growth is fertility, that
plays the most important role in changing the size and structure of population [1] and it is

PT
identified as the main cause of increasing the growth of population [2, 3]. Fertility analysis is
essential for policy makers to get guidance for population control and evaluation of family

RI
planning programs [4]. Fertility depends on the couple’s decisions, effective reproductive period,
duration of the length of birth intervals, socioeconomic, health related, and emotional factors.

SC
The intrinsic growth rate as well as the mean generational length of any population may get
affected by the birth interval [5]. This is the main evidence that the birth interval is the key factor

U
in population growth. Furthermore, information on birth interval provides an insight into birth
spacing patterns, which affect fertility as well as maternal, infant, and childhood mortality.
AN
A birth interval is defined as the time between two successive live births [6]. The duration of
birth intervals has received attention in demography and public health research because of its
M

influential implication on fertility and hazardous impact on both maternal and child health.
Several maternal and child health problems may occur due to birth interval of less than 17
D

months or more than 5 years. Different studies have been shown that short birth interval is
associated with high risk of both mother and baby death, particularly when the birth interval is
TE

less than 18 months [7]. It is evidence that there is no universally accepted optimal birth interval
[8, 9]. Zhu et al. suggests that the first birth interval should be 18-23 months to avoid the adverse
EP

prenatal outcomes [8]. According to Clayton, the optimal interval to ensure survival through
childhood is 3 years and 9 months [10]. Martin deduced that a minimum of 2 years spacing is
C

necessary between births for the best physical and mental development, but three years spacing
would be even better [11].
AC

Generally, birth intervals are higher in Bangladesh and increasing day by day. Lengthy
breastfeeding and a long period of postpartum amenorrhea are the main factors of high
percentage of births occurring after an interval of 24 months or more. Bangladesh has been
achieved a remarkable demographic transition over the last two decades. The aim of the health
sector programs those were conducted in Bangladesh is to reach a fertility level of two births per
woman within 2016 [12].

Page 2 of 16
 ACCEPTED MANUSCRIPT

Total fertility rate significantly declined from 1970 to 2014 [12]. However, birth interval is
affected by several factors. Some women are experienced very short birth interval whereas some
too long and the rest just normal. This study intends to identify the factors affecting first, second
and third birth intervals separately and assess their impacts on birth intervals.

PT
2. Methods
Data sources

RI
The current study uses the nationally representative secondary data extracted from Bangladesh
Demographic and Health Survey (BDHS) 2014. The survey is intended to serve as a source of

SC
population and health data for policy makers and the research communities in Bangladesh. The
survey was implemented by a Bangladeshi research firm “Mitra and Associates” under the

U
authority of the National Institute for Population Research and Training. Financial and technical
supports were provided by the U.S. Agency for International Development and the ICF
AN
International of Rockville, respectively.
Sample design
M

The survey is based on a two stage stratified sample of households. In the first stage, 600
enumeration areas were selected with probability proportional to the enumeration area size, with
D

207 clusters in urban areas and 393 in rural areas. In the second stage of sampling, a systematic
sample of 30 households on average from complete list of household was selected per
TE

enumeration area to provide statistically reliable estimates of key demographic and health
variables for the country. With this design, the survey selected 18,000 residential households,
EP

which were expected to result in completed interviews with about 18,000 ever married women.
All ever married women aged 15-49 years who stayed at the selected households the night before
C

the survey were eligible for the survey.

Variables of interest
AC

The identification of determinants of birth interval is a big challenge in the real world. In rural
area there was significant relation between socio demographic variables and birth interval [13,
14]. In Iran, there was significant association between birth interval with maternal age, duration
of breast feeding, sex of child, history of still births, and regular attendance at family planning
clinics [15]. Fallahian et al. showed that the duration of breastfeeding and the method of
contraceptive used were significantly associated factors with birth intervals [16]. Mothers age at

Page 3 of 16
 ACCEPTED MANUSCRIPT

marriage is measured to be an important variable in the fertility process and it is negatively

associated with birth interval [17, 18]. Education has always been an important variable in birth
[19].
In general, birth interval depends on two successive live births. The duration between first live
birth and first cohabitation of married couple is considered first birth interval. Other intervals are

PT
calculated by the difference between two successive live births. This study considered the
approximate determinants of first birth interval as region, place of residence, mothers education,

RI
partners education, mothers age, mothers working status, contraceptive use, sex of household
head, religion, wealth index and mass media exposure. But for the other birth intervals the

SC
approximate determinants as sex of previous child, survival status of previous child, duration of
previous birth interval and all the determinants of first birth interval. The necessary and required

U
categories for demographic and socio-economic factors have been presented in Table 1.
AN
The response variable for this study is birth interval. Successful birth is considered as success
event and otherwise censored. The event time is defined as the interval between date of marriage
and a live birth or censoring. Women who gave first birth but not second birth or those who were
M

not even impregnated during the survey period are considered as censored for second birth
interval analysis and this is the same for third birth interval. Pregnancy is the time during which
D

one or more offspring develops inside a woman. Childbirth occurs just over nine lunar months,
TE

where each month is about 29½ days. Women having birth interval time bellow nine months,
sterilized women and no dejure residence of women are excluded from this study.
Cox proportional hazard model and its extension
EP

In general, survival analysis examines the relationship between survival status and covariates.
Cox proportional hazard model is the most important model for survival analysis because it
C

investigates the effects of the covariates on survival [20].

AC

Let Yi denotes the observed time (either censoring time or event time) for subject i . Let C i be the

indicator that the time corresponds to an event (i.e. if C i = 1 the event occurred and if C i = 0 the

{ }
time is a censoring time). Let X i = X i1 , X i 2 ,..., X ip be the realized values of the covariates for

subject, i . The mathematical form of hazard function for the Cox proportional hazard model is:
(
h(t / X i ) = h0 (t ) exp β1 X i1 + β2 X i 2 + ... + β p X ip )

Page 4 of 16
 ACCEPTED MANUSCRIPT

(
Or, h(t / X i ) = h0 (t )exp X i β )
This expression gives the hazard rate at time t for the subject i with covariate vector
(explanatory variables) X i .
The usual Cox proportional hazard model requires time-to-event data to be independent [21].

PT
Frailty is an unobserved random proportionality factor that changes the hazard function of an
individual or a group of related individuals. In Cox proportional hazard model, it is assumed that

RI
the correlations are due to the unobservable group specific covariates and one approach to adjust
all the unobservable covariates is to include a non-negative valued random term, known as

SC
frailty. Such extension of Cox proportional hazard model is known as frailty model. Frailty
model is becoming more and more popular in the area of survival analysis and commonly used
for analyzing correlated time-to-event data [22].

U
One key problem in frailty model is choice of the frailty distribution. The most frequently used
AN
frailty distributions are gamma distribution, positive stable distribution, power variance function
(PVF) distribution, inverse Gaussian distribution, compound Poisson distribution and log-normal
distribution [23]. Among them the most common and important frailty distribution is the gamma
M

distribution. From a computational and analytical point of view the gamma distribution fits very
well to failure data, because it is easy to derive the closed form expressions of survival, density
D

and the hazard function [23].

TE

Statistical analysis
Since this study includes censored data, the survival analysis methods have been used to assess
the significant impacts of different variables on the birth interval that leads to visualize the
EP

dominant factors affecting fertility in Bangladesh. Descriptive statistics served to illustrate the
general characteristics of the variables. Kaplan-Meier estimator is used to find out the estimated
C

mean with standard error and median value of different categories of the considered covariates
AC

and log rank test is used to test the significance of different categorical factors under this study.
The Cox proportional hazard model is used to determine the effects of various factors on first
birth interval. The extension of Cox proportional hazard model known as frailty model which is
used to determine the effects of various factors on second and third birth intervals because of
heterogeneity in birth intervals due to mother. The exponent of the coefficients of Cox
proportional hazard model and frailty model is hazard ratio and have been used to calculate the
factors change in the hazard rate associated with the changes of covariates. In this study p-value

Page 5 of 16
 ACCEPTED MANUSCRIPT

less than 0.05 is considered to be statistically significant. All statistical analyses are performed
by SPSS version 15.0 and R.3.3.3.
3. Results
There were 13370, 11952 and 8871 women in this study among whom 92.40%, 76.80% and
61.60% had experienced of live birth and the rest 7.60%, 23.20% and 38.40% were censored for

PT
first, second and third birth, respectively. The average (±SD) age of the ever married women for
first, second and third births are 19.10 (±03.89), 23.13 (±05.12) and 27.78 (±06.50) years,

RI
respectively. For the ever married women the estimated average (±SE) duration of first, second
and third births are 39.05(±0.47), 76.38 (±1.13) and 135.27(±1.93) months, respectively, but the

SC
estimated median duration of births for first, second and third births are 26, 46 and 58 months,
respectively.

U
Table 2 provides descriptive statistics of the selected covariates for different birth intervals. The
result shows that the percentage of families increased with an increasing number of birth in
AN
Rangpur, but decreased in Dhaka division. Among the ever married women, 64.61% gives at
least one birth live in rural area. The percentages of rural people are increasing with respect to
M

number of birth compared to urban people. The percentage of illiterate and primary educated
mothers’ birth significantly increased from 23.64% to 30.54% and 28.15% to 31.60% but
D

secondary and higher educated mothers’ birth significantly decreased from 37.94% to 31.45%
and 10.28% to 6.41% for first birth to third birth respectively. The same scenario is discernible
TE

for partners education. In this study the majorities (88.37%) of the household head are male, and
90.30% are Muslim. For some women whose families were not connected with any mass media
EP

and their numbers of birth increased from 39.45% to 51.62%. The births increased from 35.75%
to 38.77% in poor family but this scenario is different for rich family and the number of births
C

decreased from 44.38% to 41.24%. Around 74.24% women had first birth at the age of less than
21 years, 59.26% women had second birth and also 65.02% women had third birth at the age 21-
AC

30 years. Overall, both sexes of the child are equal for each birth and for all births the survival
status of child is around 92%. In case of duration of the birth interval, maximum 45.65% have
less than 24 months for first birth, 39.57% had 24-48 months for second birth, and 47.58% had
more than 48 months for third birth.

Page 6 of 16
 ACCEPTED MANUSCRIPT

Table 2 also shows mean and median duration of birth according to the different covariates. With
the rising of wealth index, mothers and their partners education level, the birth interval
significantly increased for two or more births, but it had no effect on first birth.
Table 3 presents the different values of hazard ratio of Cox proportional hazard model and frailty
model. Rural women are significantly more likely to have experienced first, second and third

PT
birth intervals in comparison to urban women.
Highly educated mothers are significantly more likely to have experienced live birth in case of

RI
first birth interval (HR: 1.86; C.I.: 1.70-2.03) and significantly less likely to have experienced
live birth in case of third birth interval (HR: 0.47; C.I.: 0.39-0.58) if compared to others. More

SC
generally, mothers with primary, secondary and higher education are less likely to experience the
live birth than uneducated mother for second and third birth interval. Partners education (primary

U
and secondary level) is a protective factor for the next live birth for second and third birth
interval. Higher education of partners is more likely to experience to the live birth in case of first,
AN
second and third birth intervals in comparison to uneducated partners.
Female household head is more likely to experience the next live birth in comparison to male
M

household head for first and second birth interval but less likely to experience the next live birth
and significantly protective factor for third birth intervals (HR: 0.86; C.I.: 0.78-0.95). Mothers
D

working status for first, second and third birth interval is less likely to the experience in
comparison to non-working women and it is also a significantly protective factor for third birth
TE

intervals (HR: 0.94; C.I.: 0.88-1.00). Mass media plays an important role in protecting different
birth intervals. Non-Muslim family is also significantly more likely to experience the next live
EP

birth in comparison to Muslim family for first birth interval (HR: 1.12; C.I.: 1.06-1.19) but less
likely to experience the next live birth for third birth interval (HR: 0.88; C.I.: 0.79-0.97).
C

Contraceptive use is a significant protective factor for third birth interval (HR: 0.90; C.I.: 0.85-
0.96) and contraceptive use for first and second birth intervals are significantly more practiced
AC

compared to the non-users of contraceptive. Wealth index (rich and middle class family) is
approximately equally likely to the experience in comparison to poor family. In case of first birth
interval, previous birth interval is not used. But women with more than 48 months birth interval
are significantly more likely to experience the next live birth for second birth interval (HR: 1.92;
C.I.: 1.79-2.06).

Page 7 of 16
 ACCEPTED MANUSCRIPT

4. Discussion
As far as we know, there is not yet any study done on different birth intervals using BDHS 2014
data. To analyze the birth intervals, different statistical techniques and models are developed and
carried out. Therefore, there is no scope for compares the developed statistical techniques and
models in the present study with those reported in other studies. But it is important to compare

PT
this study with studies carried out in other countries.
Bangladesh is the world’s eighth most populous country with 160 million people [24]. The

RI
controlling of population growth encourages the people to longer births intervals [25]. The
estimated average duration of first two live births are 39 and 76 months, however, it is higher

SC
than the average child spacing period of Saudi Arabian children [14, 26]. This study also found
that the estimated median intervals of first two live births are 26 and 46 months but the median

U
birth interval of 55 developing countries was about 32 months [27]. Higher education of mother
is a significant protective factors for the third birth interval in comparison to first and second
AN
birth interval but Singh et al., Rajaram et al. and Gandotra at al. found that higher education of
woman was a protective factor for first birth interval [28, 29, 30]. Mass media is a protective
M

impact on higher order birth interval. This finding is supported by Singh et al. and Gandotra et al.
[28, 30].
D

Findings of this study indicate that without any doubt the mothers working status is a more
TE

significant protective factor for the first and third birth intervals. This finding is supported by
Singh et al. but not by Trussell et al. and Richter et al., who identified that employed women
were significantly less likely to go for the next birth [28, 31, 32]. The covariates contraceptive
EP

use is a significant protective factor for third birth interval but not protective factor for first and
second birth interval. The result of third birth interval is supported by Singh et al. [28]. Mothers
C

in rural area are not protective for all birth intervals and this result is supported by Singh et al,
AC

and Swenson & Thang for first and second birth interval [28, 33]. Mothers age is a significant
protective factor for all birth intervals and is supported by Swenson & Thang [33]. Rich families
are also more protective for third birth interval and this is supported by Yohannes et al. [1].
However, previous female child is not a protective factor for the second and third birth intervals.
The present study does not indicate the likelihood of it being a protective factor at any birth
interval. Survival status of previous child emerged as a significant protective factor for the
second and third birth intervals. This finding is in line with many other studies like Oheneba

Page 8 of 16
 ACCEPTED MANUSCRIPT

Sakyi & Heaton, Ojha and Palloni & Hantamala [34, 35, 36]. If the number of birth increases,
mothers education (primary, secondary and higher education), partners education (primary and
secondary) and contraceptive use also decrease.

PT
5. Conclusion
In this study, all analysis has been used to investigate the association and figure out the potential

RI
determinants of first, second and third birth interval in Bangladesh using BDHS-2014 data. This
study shows that some important covariates, which have been used in the models and varied

SC
among the birth intervals. Using log rank test we have found that the covariates division, place of
residence, mothers education, partners education, contraceptive use, mothers age, sex of the

U
previous child, duration of previous birth are significantly associated with all considered number
of births. The covariates place of residence, contraceptive use status and mothers age have
AN
significant impact on all birth intervals. Mothers education, sex of household head, mothers
working status, and religion are significant factors for first and third birth intervals. Previous sex
M

of the child, previous survival status of child and previous duration of birth are also significant
factors for second and third birth intervals. However, partners education and wealth index have
D

no significant impact on any birth intervals. This study emphasizes the importance of increasing
the partners education, regional and religion studies, role of household head and mass media
TE

exposure, and awareness of contraceptive use.

EP

Acknowledgement
Authors are profound acknowledge to Demographic and Health Surveys (DHS) Program to
C

provide data and also authors are debated to the Statistics Discipline, Khulna University where
this study is conducted. Authors are confessing uncountable debts to authors of different articles
AC

are used in this study.

Ethics Approval
We have used secondary dataset taken from the Demographic and Health Surveys (DHS)
Program website (https://dhsprogram.com/data/). No ethics approval is required for this dataset.
Funding

Page 9 of 16
 ACCEPTED MANUSCRIPT

This research did not receive any specific grant from funding agencies in the public, commercial,
or not-for-profit sectors.
Competing interests
None declared
Claiming interest

PT
Authors are not interested to face any claim about their study.

RI
Author’s contributions
Benojir Ahammed have designed the study and involved in data analysis and interpreted the

SC
result of the study and finally drafted the main manuscript. Md. Rasel Kabir and Md. Menhazul
Abedin have contributed in statistical analyses and provided overall help to prepare the

U
manuscript. All authors have read and approved the final version of the manuscript.
AN
References
1. Yohannes S, Wondafrash M, Abera M, Girma E. Duration and determinants of birth
M

interval among women of child bearing age in Southern Ethiopia. BMC Pregnancy and
Childbirth 2011; 11: 1-6.
D

2. Ayanaw A. Proximate determinants of birth interval length in Amhara region: the case of
TE

Fagita Lekoma district, Awizone. Addis Ababa, Ethiopia; 2008.

3. Yohannis F, Yemane B, Alemayehu W. Differentials of fertility in rural Butajira. Ethiop
J Health Dev 2003; 17:17-25.
EP

4. Kamal A, Pervaiz MK (2012). Determinants of Higher Order Birth Intervals in Pakistan.

Journal of Statistics 2012; 19: 54-82.
C

5. Srinivasan K. Birth Interval Analysis in Fertility Surveys. International Statistical

AC

Institute, Netherlands: Voorburg; 1980, p49.

6. Ethiopia Central Statistical Agency and ORC Macro. Ethiopia Demographic and Heath
Survey 2005, https://dhsprogram.com/pubs/pdf/FR179/FR179[23June2011].pdf ; 2006.
7. Fotso JC, Cleland J, Mberu B, Mutua M, Elungata P. Birth spacing and child mortality:
an analysis of prospective data from the Nairobi urban health and demographic
surveillance system. J Biosoc Sci. 2013; 45: 779–798.

Page 10 of 16
 ACCEPTED MANUSCRIPT

8. Zhu BP, Rolfs RT, Nangle BE, Horan JM. Effect of the interval between pregnancies on
perinatal outcomes. New England journal of medicine 1999; 340: 589-594.
9. Fallahzadeh H, Farajpour Z, Emam Z. Duration and determinants of birth interval in
Yazd, Iran: a population study. Iran J Reprod Med 2013; 11: 379–384.
10. Clayton DG. Some odds ratio statistics for the analysis of ordered categorical data.

PT
Biometrika 1974; 61: 525-531.
11. Martin EC. A study of the effect of birth interval on the development of 9-year-

RI
old schoolchildren in Singapore. J Trop Pediatr Environ Child Health 1979, 25:46-76.
12. Bangladesh demographic and health survey 2014. National Institute of Population

SC
Research and Training (NIPORT), Mitra and Associates, and ICF International,
https://dhsprogram.com/pubs/pdf/FR311/FR311.pdf; 2016.

U
13. Al-Nahedh NN. The effect of socio demographic variables on child spacing in rural Saudi
Arabia. East Mediterr Health J 1999; 5: 136-141.
AN
14. Bella H, Al-Almaie SM. Do children born before and after adequate birth intervals
do better at school? J Trop Pediatr 2005; 51: 265-270.
M

15. Hajian-Tilaki KO, Asnafi N, Aliakbarnia-Omrani F. The Patterns and determinants

of birth intervals in multiparous women in Babol, Northern Iran. Southeast Asian J
D

Trop Med Public Health 2009; 40: 852-860.

16. Fallahian M, Kazemnegat A, Ebrahimi N. Determinant of short birth interval. J
TE

Behboud Kermanshah Med Sci Univ Iran 1993; 18: 35-48.

17. Clegg E. Starting, spacing and stopping in the reproductive histories of outer Hebridean
EP

families. Journal of biosocial science 2001; 33: 405-426.

18. Nahar L, Rahman M. Changes in socio-economic differentials of age at marriage and first
C

birth during 1983-1994 in Matlab Bangladesh. Demography India 2006; 35: 1-14.
19. Shayan Z, Ayatollahi SMT, Zare N, Moradi F. Prognostic factors of first birth interval
AC

using the parametric survival models. Iran J Reprod Med 2014, 12: 125-130.
20. Cox DR. Regression Models and Life Tables. Journal of the Royal Statistical Society B
1972; 34: 187-220.
21. Singh R and Mukhopadhyay K. Survival analysis in clinical trials: Basics and must know
areas. Perspect Clin Res 2011; 2: 145–148.

Page 11 of 16
 ACCEPTED MANUSCRIPT

22. Mahmood S, Zainab B and Latif AHMM. Frailty modeling for clustered survival data: an
application to birth interval in Bangladesh. Journal of Applied Statistics 2013; 40: 2670-
2680.
23. Hanagal DD and Sharma R. Comparison of Frailty Models for Acute Leukemia Data
under Gompertz Baseline Distribution. Communications in Statistics-Theory and

PT
Methods 2015, 44: 1338-1350.
24. Bhattacharjee R, and Hasan MA. The activities of Bangladesh water development board

RI
on the social environment of Bhadrabila union, Narail sadar upazila, Bangladesh. Asian
J. Med. Biol. Res 2016; 2: 297-303.

SC
25. Abdel-Fattah M, Hifnawy T, Said TIEI, Moharam MM, Mahmoud MA. Determinants of
birth spacing among Saudi women. J Family Community Med. 2007; 14: 103–111.

U
26. Madani KA, Khashoggi RH, Al-Nowaisser AA, Nasrat HA, Khalil MH. Lactation
amenorrhea in Saudi women. J Epidemiol Community Health 1994; 48: 286-289.
AN
27. Rasheed P, Al-Dabal BK. Birth interval: perceptions and practices among urban based
Saudi Arabian women. East Mediterr Health J 2007; 13: 881-892.
M

28. Singh R, Tripathi V, Kalaivani M, Singh K, Dwivedi SN. Determinants of birth intervals
in Tamil Nadu in India: developing Cox hazard models with validations and predictions.
D

Revista Colombiana de Estadística 2012, 35: 289-307.

29. Rajaram S, Rao S, Pandey A. Birth interval dynamics in Goa: A parity specific analysis,
TE

Demography 1994, 23: 67–81.

30. Gandotra M, Retherford R, Pandey A, Luther N, Mishra V. (1998). Fertility in India.
EP

National Family Health Survey Subject Reports 9, Mumbai: International Institute for
Population Sciences, Honolulu; 1998.
C

31. Trussell J, Martin L, Fledman R, Palmore J, Concepcion M, Bakar, DA. Determinants of

birth interval length in the Phillipines, Malaysia and Indonesia: A hazard model analysis.
AC

Demography 1985; 22:145-168.

32. Richter K, Podhisita C, Chamratrithirong A, Soonthorndhada K. (1994). The impact of
child care on fertility in urban Thailand. Demography 1994, 31: 651–662.
33. Swenson I, Thang NM. Determinants of birth intervals in Vietnam: A hazard model
analysis. Journal of Tropical pediatrics 1993; 39: 163–167.

Page 12 of 16
 ACCEPTED MANUSCRIPT

34. Ojha, A. The effect of sex preference on fertility in selected states of India. The Journal
of Family Welfare 1998; 44: 42–48.
35. Oheneba-Sakyi, Y. & Heaton, TB. Effects of socio-demographic variables on birth
intervals in Ghana. Journal of Comparative Family Studies 1993; 24: 113–135.
36. Palloni, A. & Hantamala, R. The e ects of infant mortality on fertility revisited: New

PT
evidence from Latin America. Demography 1999; 36:41–75.

RI
Table 1: The category of demographic and socio-economic characteristics with identification code.
Demographic and socio-economic Categories (code)
characteristics

SC
Region/Division Barisal (1), Chittagong (2), Dhaka (3), Khulna (4), Rajshahi (5),
Rangpur (6), Sylhet (7)
Place of residence Urban (1), Rural (2)
Mothers education No education (0), Primary (1), Secondary (2), Higher education (3)

U
Partners education No education (0), Primary (1), Secondary (2), Higher education (3)
Mothers age ≤ 20 years (1), 21-30 years (2), ≥31 years (3)
AN
Mothers working status No (0), Yes (1)
Contraceptive use status No (0), Yes (1)
Sex of household head Male (1), Female (2)
M

Religion Muslim (0), Non-Muslim (1)

Wealth index Poor (0), Middle (1), Rich (2)
Mass media exposure No (0), Yes (1)
Sex of previous child Boy (1), Girl (2)
D

Survival status of previous child Dead (0), Alive(1)

Previous birth interval <24 months (1), 24-48 months (2), >48 months (3)
TE
C EP
AC

Page 13 of 16
 ACCEPTED MANUSCRIPT

Table 2: Estimated mean and median duration of first, second and third birth interval according to different categorical covariates.
Mean and median duration of first Mean and median duration of second Mean and median duration of third
birth interval birth interval birth interval
Number Mean Number Mean Mean
Med Sig. Med Sig. Number (%) Med Sig.
Factor (%) ±SE (%) ±SE ±SE

PT
Region/Division
Barisal 1571 (11.75%) 37.55±1.05 28 1387 (11.60%) 69.68±2.77 46 1048 (11.81%) 107.46±4.35 53
Chittagong 2084 (15.59%) 32.35±0.77 22 1853 (15.50%) 61.72±2.39 40 1435 (16.18%) 87.54±3.14 46
Dhaka 2353 (17.60%) 39.75±1.04 27 2083 (17.43%) 76.07±2.37 50 1484 (16.73%) 140.97±4.54 68

RI
Khulna 2000 (14.96%) 41.88±1.28 28 <0.001 1815 (15.19%) 86.72±2.82 54 <0.001 1326 (14.95%) 169.54±5.35 81 <0.001
Rajshahi 1906 (14.26%) 43.21±1.45 28 1700 (14.22%) 85.31±2.97 55 1210 (13.64%) 156.93±±5.03 78
Rangpur 1930 (14.44%) 37.42±0.92 27 1769 (14.80%) 76.39±2.86 46 1326 (14.95%) 138.94±4.51 66

SC
Sylhet 1526 (11.41%) 38.92±1.50 23 1345 (11.25%) 54.60±2.12 36 1042 (11.75%) 72.53±3.37 39
Place of residence
Urban 4731 (35.39%) 40.39±0.81 27 4162 (34.82%) 90.52±2.22 51 2931 (33.04%) 149.79±3.24 72
Rural 8639 (64.61%) 38.24±0.56 26
<0.001 7790 (65.18%) 68.05±1.21 43
<0.001
5940 (66.96%) 124.06±2.27 53
<0.001

U
Mothers educational
No education 3160 (23.64%) 45.50±1.01 30 3023 (25.29%) 63.52±1.73 36 2709 (30.54%) 101.14±2.69 42

AN
Primary 3764 (28.15%) 37.21±0.76 25 3465 (28.99%) 62.94±1.51 41 2803 (31.60%) 111.58±2.71 52
<0.001 <0.001 <0.001
Secondary 5072 (37.94%) 35.65±0.76 25 4370 (36.56%) 84.70±2.06 55 2790 (31.45%) 159.47±3.66 81
Higher 1374 (10.28%) 37.96±0.95 29 1094 (9.15%) 105.29±4.50 66 569 (6.41%) 240.91±8.55 .

M
Partners education
No education 3727 (27.88%) 41.55±0.82 28 3528 (29.52%) 62.47±1.56 38 3010 (33.93%) 102.78±2.59 47
Primary 3549 (26.54%) 36.07±0.84 24 3185 (26.65%) 64.46±1.66 43 2409 (27.16%) 114.39±3.15 52
<0.001 <0.001 <0.001
Secondary 3938 (29.45%) 38.39±0.89 26 3390 (28.36%) 84.68±2.37 51 2283 (25.74%) 157.90±4.15 70

D
Higher 2156 (16.13%) 39.51±1.15 28 1849 (15.47%) 99.47±3.15 60 1169 (13.18%) 190.61±5.54 102
Sex of household head

TE
Male 11815(88.37%) 39.03±0.51 26 10532 (88.12%) 74.14±1.17 46 7810 (88.04%) 133.77±2.05 58
0.122 0.022 0.137
Female 1555 (11.63%) 39.21±1.14 27 1420 (11.88%) 88.57±3.58 46 1061 (11.96%) 133.05±4.79 59
Working status
No 8996 (67.28%) 37.98±0.55 26 7880 (65.93%) 72.81±1.31 46 5673 (63.95%) 124.75±2.22 57
EP
<0.001 0.092 0.001
Yes 4374 (32.72%) 40.90±0.81 28 4072 (34.07%) 80.83±1.99 46 3198 (36.05%) 144.55±3.23 61
Mass Media exposure (TV, Radio, etc.)
No 6642(49.68%) 39.45±0.66 26 5968 (49.93%) 68.37±1.44 41 4579 (51.62%) 110.23±2.30 50
0.498 <0.001 <0.001
C

Yes 6728(50.32%) 38.55±0.65 26 5984 (50.07%) 82.44±1.62 51 4292 (48.38%) 157.71±2.95 71

Religion
AC

Muslim 12073(90.30%) 39.09±0.49 26 10792 (90.29%) 74.92±1.17 46 8040 (90.63%) 131.35±1.99 57

0.768 <0.001 <0.001
Non-Muslim 1297(9.70%) 37.64±1.19 26 1160 (9.71%) 86.94±3.84 51 831 (9.37%) 153.88±5.74 74
Wealth index
Poor 4780(35.75%) 39.31±0.73 27 4359 (36.47%) 63.90±1.50 40 3439 (38.77%) 104.82±2.58 48
Middle 2657(19.87%) 37.89±1.03 26 <0.001 2362 (19.76%) 70.56±2.10 45 <0.001 1774 (20.00%) 115.33±3.62 54 <0.001
Rich 5933(44.38%) 39.14±0.71 26 5231 (43.77%) 87.78±1.95 52 3658 (41.24%) 166.64±3.26 77

Page 14 of 16
 ACCEPTED MANUSCRIPT

Table 2 continue……………….
Contraceptive use status
No 5451(40.77%) 52.04±1.17 24 <0.001 4442 (37.17%) 96.93±2.23 46 <0.001 3122 (35.19%) 127.73±3.01 47 <0.001
Yes 7919(59.23%) 31.58±0.28 30 7510 (62.83%) 62.48±1.04 46 5749 (64.81%) 132.42±2.33 64
Mothers age

PT
≤20 years 9927 (74.24) 27.35±0.17 23 3946 (33.02%) 27.35±0.17 23 NA NA
21-30 years 3219 (24.08) 56.22±0.75 45 <0.001 7083 (59.26%) 56.22±0.75 45 <0.001 NA NA NA
≥31 years 224 (01.68) 248.54±11.42 234 923 (7.72%) 248.54±11.42 234 NA NA
Duration of birth

RI
<24 Months 6104 (45.65%) 15.61±0.06 15 2517 (21.06%) 18.35±0.09 19 1540 (17.36%) 18.70±0.11 20
24-48 Months 4738 (35.44%) 33.95±0.11 33 <0.001 4730 (39.57%) 35.75±0.11 36 <0.001 3110 (35.06%) 36.23±0.14 36 <0.001
>48 Months 2528 (18.91) 96.89±1.78 70 4705 (39.37%) 129.30±2.19 78 4221(47.58%) 215.85±2.87 123

SC
Sex of previous child
Male 6110 (51.12%) 81.28±1.67 48 4465 (50.33%) 148.12±2.82 64
<0.001 <0.001
Female 5842 (48.88%) 70.39±1.45 44 4406 (49.67%) 115.95±2.36 53
Survival status of previous child

U
Died 1174 (9.82%) 38.07±1.68 25 659 (7.43%) 48.21±2.86 28
<0.001 <0.001
Alive 10778 (90.18%) 80.77±1.24 49 8212 (92.57%) 142.44±2.05 62

AN
Duration of previous birth
<24 Months 5482 (45.87%) 69.19±1.52 46 1540 (17.36%) 18.70±0.11 20
24-48 Months 4301 (35.99%) 72.99±1.64 46 <0.001 3110 (35.06%) 36.23±0.14 36 <0.001
>48 Months 2169 (18.15%) 86.90±2.60 47 4221 (47.58%) 215.85±2.87 123

M
Overall 39.05±0.47 26 76.38±1.13 46 135.27±1.93 58
Med, Median; SE, Standard Error; Sig, Significance.

D
TE
C EP
AC

Page 15 of 16
 ACCEPTED MANUSCRIPT

Table 3: Estimates of Hazard Ratio (95%CI) and p-value of Cox proportional hazard model for first birth interval and Frailty model for
second and third birth intervals according to different categorical variables.
First birth interval Second birth interval Third birth interval
HR (95% CI ) p-value HR (95% CI ) p-value HR (95% CI) p-value
Region/Division
Barisal 1.00 1.00 1.00
Chittagong 1.30 (1.21,1.39) <0.001 1.38 (1.26,1.51) <0.001 1.16 (1.04,1.29) 0.006
Dhaka 1.07 (1.00,1.15) 0.055 0.93 (0.85,1.02) 0.100 0.85 (0.76,0.95) 0.003
Khulna 0.96 (0.89,1.03) 0.238 0.77 (0.70,0.85) <0.001 0.74 (0.66,0.83) <0.001

PT
Rajshahi 0.97 (0.90,1.04) 0.342 0.77 (0.70,0.84) <0.001 0.72 (0.64,0.81) <0.001
Rangpur 1.00 (0.93,1.07) 0.983 0.87 (0.80,0.96) 0.003 0.72 (0.65,0.81) <0.001
Sylhet 1.47 (1.36,1.59) <0.001 1.62 (1.47,1.79) <0.001 1.41 (1.26,1.58) <0.001
Place of residence

RI
Urban 1.00 1.00 1.00
Rural 1.05 (1.01,1.09) 0.024 1.10 (1.04,1.16) 0.001 1.10 (1.03,1.18) 0.005
Mothers education
No education 1.00 1.00 1.00

SC
Primary 1.16 (1.11,1.22) <0.001 0.89 (0.83,0.95) <0.001 0.75 (0.70,0.81) <0.001
Secondary 1.32 (1.25,1.40) <0.001 0.72 (0.67,0.78) <0.001 0.53 (0.48,0.58) <0.001
Higher 1.86 (1.70,2.03) <0.001 0.89 (0.79,1.01) 0.070 0.47 (0.39,0.58) <0.001
Partners education
No education 1.00 1.00 1.00

U
Primary 1.10 (1.04,1.15) <0.001 0.95 (0.89,1.02) 0.140 0.98 (0.91,1.06) 0640
Secondary 1.05 (0.99,1.11) 0.112 0.92 (0.86,0.99) 0.024 0.99 (0.90,1.08) 0.780
Higher 1.07 (0.99,1.15) 0.076 1.10 (0.99,1.21) 0.066 1.15 (1.01,1.32) 0.033
AN
Sex of household head
Male 1.00 1.00 1.00
Female 1.07 (1.01,1.13) 0.033 1.03 (0.95,1.11) 0.520 0.86 (0.78,0.95) 0.002
Mothers working status
No 1.00 1.00 1.00
M

Yes 0.95 (0.91,0.99) 0.007 0.95 (0.91,1.00) 0.052 0.94 (0.88,1.00) 0.035
Mass media exposure
No 1.00 1.00 1.00
Yes 0.99 (0.94,1.04) 0.622 0.98 (0.91,1.04) 0.400 0.97 (0.89,1.04) 0.380
D

Religion
Muslim 1.00 1.00 1.00
Non-Muslim 1.12 (1.06,1.19) <0.001 1.05 (0.97,1.14) 0.260 0.88 (0.79,0.97) 0.014
TE

Wealth index
Poor 1.00 1.00 1.00
Middle 1.00 (0.95,1.06) 0.941 0.98 (0.92,1.06) 0.660 1.04 (0.96,1.14) 0.330
Rich 1.00 (0.94,1.06) 0.921 1.01 (0.93,1.09) 0.890 1.02 (0.92,1.12) 0.720
Contraceptive use status
EP

No 1.00 1.00 1.00

Yes 1.36 (1.31,1.41) <0.001 1.27 (1.21,1.34) <0.001 0.90 (0.85,0.96) 0.002
Mothers age
≤20 years 1.00 1.00 1.00
C

21-30 years 0.31 (0.30,0.33) <0.001 0.20 (0.19,0.21) <0.001 0.22 (0.20,0.25) <0.001
≥31 years 0.04 (0.03,0.05) <0.001 0.02 (0.01,0.03) <0.001 0.02 (0.01,0.03) <0.001
Sex of previous child
AC

Male 1.00 1.00

Female 1.15 (1.10,1.20) <0.001 1.24 ( 1.17,1.32) <0.001
Survival status of previous child
Died 1.00 1.00
Alive 0.41 (0.38,0.45) <0.001 0.50 (0.45,0.55) <0.001
Previous birth interval
<24 months 1.00 1.00
24-48 months 1.23 (1.17,1.30) <0.001 1.06 (0.99,1.14) 0.120
>48 months 1.92 (1.79,2.06) <0.001 0.99 (0.91,1.08) 0.840
HR, Hazard Ratio; 95% CI, 95% confidence interval.

Page 16 of 16