Postpartum amenorrhoea among Manipuri women: a survival analysis.

Abstract

Among the three major components of a closed birth interval, waiting time to conception can somehow be managed with effective contraceptives while gestation is universally constant in its duration; the duration of postpartum amenorrhoea (PPA) varies in complex nature. The present study aimed to investigate the proximate factors influencing the duration of PPA. A community-based, cross-sectional study was conducted in four valley districts of Manipur, India, during 1 August-31 December 2009, to analyze the differentials and determinants of duration of PPA, applying the survival analysis technique. In total, 1,225 ever-married women were selected through two-stage cluster sampling. The median duration of PPA was 5.7 months. Among the 11 explanatory variables of interest, only three variables--place of residence (p<0.05), infant mortality from preceding pregnancy (p<0.01), and duration of breastfeeding (p<0.01)--had a significant effect on the duration of PPA. The findings may be used as baseline information for future researchers and maternal health policy-makers.

Full text

Page 1

©INTERNATIONAL CENTRE FOR DIARRHOEAL
DISEASE RESEARCH, BANGLADESH
Correspondence and reprint requests should be
addressed to:
Dr. N. Sanajaoba Singh
Directorate of Census Operations
Porompat, Manipur 795 005
India
Email: biostatsana@gmail.com
Postpartum Amenorrhoea among Manipuri
Women: A Survival Analysis
N. Sanajaoba Singh1, N. Sharat Singh2, and R.K. Narendra3
1Directorate of Census Operations, Porompat, Manipur, India, 2Thoubal College, Thoubal, Manipur, India, and
3Unit of Biostatistics, Regional Institute of Medical Sciences, Imphal, India
ABSTRACT
Among the three major components of a closed birth interval, waiting time to conception can somehow
be managed with effective contraceptives while gestation is universally constant in its duration; the du-
ration of postpartum amenorrhoea (PPA) varies in complex nature. The present study aimed to investi-
gate the proximate factors influencing the duration of PPA. A community-based, cross-sectional study was
conducted in four valley districts of Manipur, India, during 1 August−31 December 2009, to analyze
the differentials and determinants of duration of PPA, applying the survival analysis technique. In total,
1,225 ever-married women were selected through two-stage cluster sampling. The median duration of PPA
was 5.7 months. Among the 11 explanatory variables of interest, only three variables—place of residence
(p<0.05), infant mortality from preceding pregnancy (p<0.01), and duration of breastfeeding (p<0.01)—
had a significant effect on the duration of PPA. The findings may be used as baseline information for future
researchers and maternal health policy-makers.
Key words: Breastfeeding; Community-based studies; Cross-sectional studies; Infant mortality; Parity;
Postpartum amenorrhoea; India
J HEALTH POPUL NUTR 2012 Mar;30(1):93-98
ISSN 1606-0997 | $ 5.00+0.20
INTRODUCTION
The duration of postpartum amenorrhoea (PPA) is
the length of time between the termination of preg-
nancy and the first successive ovulation following
the pregnancy during the reproductive span of a
woman. In other words, it is the temporary sterility
immediately following the termination of pregnan-
cy during which conception usually does not occur.
The attainment of first menstruation after delivery
is treated as the termination of PPA. Being a physi-
ological process associated with each conception,
the importance of PPA in regulating human fertili-
ty has been a matter of general interest for many
years because it tends to increase the average birth
interval, thereby reducing women’s fertility over
her life span, especially in societies where the use of
contraceptive methods is not widespread. There is
substantial evidence that long-term breastfeed-
ing is associated with a longer anovulatory period
leading to prolonged PPA. It also reduces fertility
even after menses and ovulation have returned.
Prolonged lactation suppresses the production of
certain types of hormones, thereby extending the
postpartum anovulatory period. Moreover, there
are different social norms, taboos, and other socio-
economic factors, especially in traditional societies,
that reinforce the effect of breastfeeding. How-
ever, many mechanisms by which breastfeeding
behaviour and other factors produce variability in
the length of the anovulatory period and thus, in
the resumption of menses across populations and
within a population in different social and cultur-
al groups remain either unspecified or unknown.
They are probably associated with the biological
characteristics and specific social structures and im-
plicit or explicit social norms.
The anovulatory period depends on a number of
other factors, which vary from woman to woman
in a population (1,2). PPA is associated with age of
mother, marital duration, nutritional status, and
parity (3-7). In these studies, breastfeeding is also
found to be associated with delayed return of men-
strual cycles and ovulation, reduced fecundability,
and long birth intervals. The most important con-
clusion is that long-term breastfeeding was statisti-
cally associated with a long period of PPA (6,8-11).
In the mid-1970s, some pioneering studies also
stressed that, at least at the aggregate level, when
regressing the average durations of PPA on the aver-

Page 2

Singh NS et al.
JHPN
94
Postpartum amenorrhoea among Manipuri women
age durations of breastfeeding, the two are signifi-
cantly and positively related (12,13). However, in
the absence of breastfeeding, the average amenor-
rhoeic period may last 1-3 months (11,12). Educa-
tion and family income have been inversely related
with the duration of PPA (7,10,13).
In the policy point of view, the present empirical
study was conducted in Manipur, one of the north-
eastern states of India. The study was conducted to
investigate the nature and pattern of duration of
PPA and to check the relative influences of various
sociodemographic factors on it. The findings may
be of immense values, particularly in couple’s re-
productive management, population planning,
and health policy through which an attempt may
also be made to improve the sociodemographic sta-
tus of women in Manipur. Besides, the work is of
considerable importance from the academic and
application points of view, and as such, it assumes
great significance as baseline information of repro-
ductive health.
MATERIALS AND METHODS
Data
This community-based, cross-sectional study was
conducted in four valley districts of Manipur,
namely Bishnupur, Thoubal, Imphal West, and
Imphal East, during 1 August—31 December 2009
using the two-stage cluster-sampling technique,
with 11 July as the reference date of survey. The
study subject was defined to be an eligible
woman having at least one livebirth. The pre-
tested and semi-structural type of interview schedule
was used as the tool for collecting the required in-
formation. The sample consisted of 1,225 eligible
women. The sample-size was determined based on
the findings of Sharat (14), i.e. the mean±standard
deviation (SD) of duration of closed birth interval is
21.05±18.79, under 95% degree of precision, with
5% margin of error to the mean (14). Altogether,
clusters of 45 rural villages and 35 clusters of urban
wards were randomly selected, of which 5 villages
and 7 wards, 9 villages and 10 wards, 11 villages
and 12 wards, and 20 villages and 6 wards were, res-
pectively, selected proportionally from Bishnupur,
Thoubal, Imphal West, and Imphal East districts.
Of the 1,225 eligible women, 180, 316, 387, and
342 were randomly picked up from the selected
clusters of Bishnupur, Thoubal, Imphal West, and
Imphal East districts respectively.
Response variable
The duration of PPA was taken as the response vari-
able. Since ovulation itself is difficult to identify,
the reliable estimate of the end of amenorrhoea
is the return of menstruation itself. In this study,
the duration of PPA was estimated as an interval be-
tween the termination of conception and the
return of first menses. Women reporting the termi-
nation of amenorrhea before the survey date were
considered uncensored cases, and their duration
of PPA was the interval between the termination
of conception and the return of the first mense. On
the other hand, women reporting continuance of
amenorrhoea even after the survey date were con-
sidered censored cases, and their event history of
interest was the interval between the termination
of conception and the survey date. In the case of
women having more than one livebirth, the dura-
tion variable, i.e. the duration of PPA for the last
birth, was taken into consideration to control the
data recall error.
Explanatory variables
The explanatory variables or so termed as covari-
ates were demographic and socioeconomic factors.
The socioeconomic variables included the place of
residence, educational level, and family income.
Under the demographic variables, duration of
breastfeeding, age at marriage, sex of the previous
child, parity and living status of the previous child,
and use of contraceptive devices were considered.
These variables were expected to cause the variation
in the duration of PPA in view of the past findings.
It was rather difficult to explore the effects of the
explanatory variables in isolation owing to the fact
that these were purely interdependent on one an-
other. Here, the adopted tools had to quantify the
possible effects of the variables after adjusting the
effects of other variables under consideration.
Methods of analysis
As the study was confined in the censored data,
survival analysis techniques were used for statisti-
cal analysis. The Log-rank test was employed to
compare the survival experience between different
groups under study. The Cox’s proportional haz-
ard model was used for determining the effects of
breastfeeding variables and other demographic and
socioeconomic factors on the duration of PPA. The
simple form of the Cox’s proportional hazard mod-
el was given by λ (t;x)=λ0(t) exp (β/, x), where λ0(t) is
the baseline hazard function and β is a vector of un-
known regression coefficients to be estimated (15).
The ratio—λ (t;x)/λ0(t)=exp (β/,x)—represents the
‘relative risk’ of failure or so termed as the risk of
resumption of menstruation. Therefore, the hazard
function enables one to estimate the relative risks of
other groups in relation to the baseline group (ref-
erence group). When there is no covariate present
in the model, exp (β/, x) is unity. The results are

Page 3

Singh NS et al.
Postpartum amenorrhoea among Manipuri women
Volume 30 | Number 1 | March 2012
95
expressed in terms of β-coefficients, standard error
(SE), p values, and relative risk (RR: eβ) with their
95% confidence interval (CI). The value of relative
risk of a group of the study subjects greater than
unity means that the risk of resumption of men-
struation is higher for this group when compared
with the reference group while values of less than
one indicate that the risk is lower for the case group
being analyzed when compared with the baseline
group.
RESULTS
Univariate analysis
Table 1 shows the median durations of PPA accord-
ing to the specific demographic and socioeconom-
ic characteristics of the study women. The median
duration of PPA in the study subjects was 5.7
months. The median duration of PPA increased
with the increase in the duration of breastfeeding.
By the Log-rank test, the association between the
duration of breastfeeding and PPA was highly sig-
nificant (χ2=41.439, p<0.01). A negative and strong
association was found between the duration of PPA
and the age at marriage of women (p<0.01). The
median duration of PPA increased with the increase
in parity, and this variation was highly significant
(p<0.01) in the study subjects, irrespective of
other covariates. The sex of the previous child did
not influence PPA (p>0.05). The survival status of
the previous child showed a strong effect on the
duration of PPA (p<0.01). The longer duration of
PPA was found with mothers who had an alive
child (5.8 months) than who experienced a dead
child (4.2 months). The duration of PPA for the
couple who used contraceptive devices was signifi-
cantly shorter than those who had never used any
form of contraceptive devices (p<0.05). Non-users
resumed menses six months after childbirth while
users resumed menses after 5.3 months. The vari-
ation in the median duration of PPA according to
educational level of couple was a downward linear
trend, which was again highly significant (p<0.01).
Multivariate analysis
To assess the partial effects of the explanatory vari-
ables on the duration of PPA while controlling for
other covariates, multivariate proportional hazard
model analysis was performed. The results are pre-
sented in Table 2. After the adjustment of other co-
variates under study, the duration of breastfeeding
had a negative association with the risk of return to
menses (β=-0.015, p<0.01). In fact, for every one-
month increase in the duration of breastfeeding,
the risk of resumption of menses decreased by 1.5%
(RR=0.985). The survival status of the previous child
during infancy had a significant effect (p<0.01)
on the duration of PPA. The risk of resumption of
menses of women with the survival of the previous
child was 0.58 times lower than those who experi-
enced the death of the previous child during in-
fancy (RR=0.584). Among the socioeconomic vari-
ables, the place of residence had a significant effect
(p<0.05) on the risk to return of menstruation after
controlling for all other explanatory variables. The
rural women were found to be subject to a hazard
of menstruation 0.835 lower than that of urban
women (RR=0.835).
DISCUSSION
The findings showed that the median duration of
PPA was 5.7 months. The findings also highlighted
that the longer duration of breastfeeding was sig-
nificantly associated with a lower risk of menstrua-
tion. This is due to the fact that prolonged lacta-
tion suppresses the production of certain types of
hormones, thereby extending the postpartum ano-
vulatory period. In fact, longer breastfeeding has
suckling stimulation and proper prolactin secretion
by the pituitary gland. Prolactin again suppresses go-
nadotropin (GnRH), which delays the secretion of
follicle stimulation hormone (FSH) and Luteinizing
hormone (LH), resulting in longer PPA. The results
of this study are consistent with findings of studies
in different countries (8-12, 16-18). The death of the
index child during infancy was associated with the
short duration of PPA, which is also consistent with
findings of other studies (19,20). It may be possi-
ble that the death of a child during infancy cuts
short the duration of nursing, which results in
earlier resumption of menses and ovulation. The
finding of the present study is also in the same di-
rection.
The effects of age at marriage, parity, sex of the pre-
vious child, use of contraceptive devices, and parity
were not at all significant. The place of residence
had a significant relationship with the duration of
PPA. The risk of resumption of menses for urban
women was higher than rural women. The shorter
duration of PPA among urban women was perhaps
due to the fact that the educational level and em-
ployment status were higher among urban women
than among rural women. In other words, the so-
cioeconomic status of rural women was compara-
tively low due to their low level of education and
employment. In contrast, most urban employed
mothers did not get enough time to breastfeed
their children as they work outside and, thus, tend
to lactate for a shorter period and also probably
provide food supplements to their children much
earlier.

Page 4

Singh NS et al.
JHPN
96
Postpartum amenorrhoea among Manipuri women
Table 1. Univariate analysis of duration of postpartum amenorrhoea
Variable
Median of PPA
(months)
Log-rank test (χ2)
Demographic
Duration (months) of breastfeeding
<5
5-10
10-15
15-20
20-25
≥25
Age (years) at marriage of wife
<15
15-20
20-25
25-30
≥30
Parity
0
1
2
3
4+
Sex of previous child
Female
Male
Survival status of previous child
Death
Survival
Use of contraceptives
No
Yes
Socioeconomic factors
Place of residence
Urban
Rural
Educational level of husband
No schooling
Primary school
Seondary school
Higher seondary school
College and university
Educational level of wife
No schooling
Primary school
Seondary school
Higher seondary school
College and university
41.439, p<0.01
4.8
5.1
5.8
6.2
6.8
8.4
23.705, p<0.01
6.4
5.8
5.1
4.5
4.3
23.068, p<0.01
4.3
5.1
5.8
6.4
6.9
0.334, p>0.05
5.1
6.2
8.340, p<0.01
4.2
5.8
5.453, p<0.05
6.0
5.3
5.612, p<0.05
5.2
6.0
16.672, p<0.01
6.9
6.0
5.6
5.2
4.6
15.403, p<0.01
6.8
6.1
5.7
5.5
4.9
Contd.

Page 5

Singh NS et al.
Postpartum amenorrhoea among Manipuri women
Volume 30 | Number 1 | March 2012
97
Table 2. Cox’s multivariate regression analysis of duration of postpartum amenorrhoea
VariableCoefficient (β) Wald test (p value)
Demographic
Duration (months)
of breastfeeding -0.01511.703, p<0.01
Age (years) at
marriage of wife0.011
Parity-0.027
Sex of previous child-0.003
Survival status of
previous child-0.538
Use of contraceptives -0.008
Socioeconomic
Place of residence -0.180
Educational level of
husband0.004
Educational level of
wife -0.005
Family income (Rs)0.000
CI=Confidence interval
Relative risk (eβ) 95% CI for eβ
0.985 0.976-0.994
1.588, p>0.05
2.393, p>0.05
0.003, p>0.05
1.011
0.973
0.997
0.994-1.029
0.940-1.007
0.890-1.117
8.167, p<0.01
0.011, p>0.05
0.584
0.992
0.390-0.845
0.859-1.146
5.710, p<0.050.835 0.721-0.968
0.313, p>0.051.004 0.989-1.019
0.465, p>0.05
0.727, p>0.05
0.995
0.999
0.980-1.010
0.998-1.001
This finding is in agreement with findings of Srini-
vasan et al. (2), Aguirre et al. (6), and Aryal (7). De-
spite the importance of parity in the determination
of the amenorrhoeic period, we found that this
variable was not at all significant. It seems that its
effect is absorbed totally by other related variables.
This finding is contradictory with findings of other
studies (3,7,10).
The findings of this interpretative study confirm
that PPA is highly influenced by the duration of
breastfeeding. Breastfeeding is again associated
with other explanatory variables in many ways
in the study subjects due to the diverse sociocul-
tural and behavioural aspects. The policy-makers
and planners, particularly maternal health plan-
ners are thus, suggested to promote breastfeeding
practice and its associated factors to reduce fertil-
ity in the study subjects through prolonged birth
intervals.
Conclusions
The factors—breastfeeding, infant mortality, and
residence―were found to have their high impacts
on the duration of PPA. Prolonged breastfeeding
can delay the resumption of menses following a
livebirth, leading to lengthening of the subsequent
birth intervals to reduce total fertility to achieve the
national goal of replacement fertility of 2.1. Death
of infants cuts down the nursing period, resulting in
early resumption of menses and ovulation. Besides,
evidence of high risk of short PPA among urban
women might have been associated with higher ed-
ucation and employment, which are again linked
to reduce the breastfeeding period. Hence, popula-
tion scientists and health planners are needed to
improve the factors associated with the prolonged
duration of breastfeeding so that the eligible moth-
ers could extend the duration of PPA.
Table 1—Contd.
Variable
Median of PPA
(months)
Log-rank test (χ2)
Family income (Rs)
<2,000
2,000-<4,000
4,000-<6,000
6,000-<8,000
8,000-<10,000
≥10,000
Overall
PPA=Postpartum amenorrhoea
7.306, p<0.05
6.4
6.0
5.7
5.3
4.8
4.3
5.7