Duration of inter-pregnancy interval and its predictors among pregnant women in urban South Ethiopia: Cox gamma shared frailty modeling.

BACKGROUND: Short inter-pregnancy interval is a public health concern because it results in adverse perinatal outcomes such as postpartum hemorrhage, anemia, premature birth, low birth weight, and perinatal deaths. Although it is critical to understand the factors that contribute to short inter-pregnancy interval to reduce the risk of these negative outcomes, adequate evidence about the factors in the urban context is lacking. Therefore, we aimed to assess the duration of the inter-pregnancy interval and its predictors among pregnant women in urban South Ethiopia.
METHODS: A community-based retrospective follow-up study was conducted among 2171 pregnant women in five geographically diverse urban settings in South Ethiopia. For the analysis, a Cox gamma shared frailty (random-effect) model was used. Adjusted hazard ratio (AHR) with a 95% CI was used to assess significant predictors. The median hazard ratio (MHR) used to report clustering effect.
RESULTS: The median duration of the inter-pregnancy interval was 22 months, 95% CI (21, 23), with an inter-quartile range of 14 months. Maternal age ≥30 years [AHR = 0.75, 95% CI: 0.58, 0.97], having no formal education [AHR = 0.60, 95% CI: 0.46, 0.78], contraceptive non-use [AHR = 2.27, 95% CI: 1.94, 2.66], breastfeeding for <24 months [AHR = 4.92, 95% CI: 3.95, 6.12], death of recent child [AHR = 2.90, 95% CI: 1.41, 5.97], plan pregnancy within 24 months [AHR = 1.72, 95% CI: 1.26, 2.35], lack of discussion with husband [AHR = 1.33, 95% CI: 1.10, 1.60] and lack of husband encouragement about pregnancy spacing [AHR = 1.25, 95% CI: 1.05, 1.48] were predictors of short inter-pregnancy interval. Adjusting for predictors, the median increase in the hazard of short inter-pregnancy interval in a cluster with higher short inter-pregnancy interval is 30% [MHR = 1.30, 95% CI: 1.11, 1.43] than lower cluster.
CONCLUSIONS: In the study settings, the duration of the inter-pregnancy interval was shorter than the World Health Organization recommendation. There is a need to improve contraceptive use and breastfeeding duration to maximize the inter-pregnancy interval. Men's involvement in reproductive health services and advocacy for women's reproductive decision-making autonomy are fundamental. The contextual disparities in the inter-pregnancy interval suggests further study and interventions.

Introduction

Inter-pregnancy interval (IPI), also known as birth to pregnancy interval, is defined as the time elapsed from a live birth to subsequent conception or a woman’s last menstrual period (LMP) [1]. The World Health Organization (WHO) recommended at least 24 months between a live birth and subsequent conception, and IPI less than 24 months is generally considered short [1, 2].

Every day in 2017, approximately 810 maternal deaths occur around the world, with 94% occurring in low and middle-income countries [3]. Ethiopia, with a Maternal Mortality Ratio (MMR) of 412 per 100,000 live births, is one of the countries in Sub-Saharan Africa that contributes to unacceptably high level of maternal mortality (295,000) worldwide [3, 4]. Women in developing countries, including Ethiopia, have many more pregnancies and fertility rate than in developed countries [5]. Pregnancies that are many and IPIs that are short and long are public health concerns that have drawn the attention of policymakers and programs due to the negative outcomes associated with them [1]. To be more specific, both short and long IPIs are associated with poor maternal and neonatal outcomes [1, 6]. For instance, IPI <6, <18 and <24 months are associated with an increased risk of prematurity, low birth weight and small for gestational age, which are linked to neonatal death [7-9]. IPI <18 months is associated with an increased risk of adverse maternal outcomes such as anemia, postpartum hemorrhage, and pre-eclampsia, which are associated with maternal death [7, 10]. Longer IPI (>59 months) is associated with pre-eclampsia [11]. By lowering these risks, adequate IPI improves maternal and child survival.

In Ethiopia, contrary to the WHO recommendation, a sizeable proportion of women were not spacing their pregnancies adequately [1, 4]. Still, more than half of second and higher pregnancies occur within a shorter interval than recommended by WHO (at least 24 months) [4]. More specifically, about 47.9% of women in urban areas have a short IPI <24 months [2]. As in many developing countries, the total fertility rate (TFR) in Ethiopia has declined steadily [5], and it has taken 20 years to decline from 5.5 in 2000 to 4.2 in 2019; on average, one child per woman [12]. The current TFR of 4.2 is still high and no more different from the overall TFR in Sub-Saharan Africa (4.6) [5]. A high TFR and a short IPI may result in suffering and poor maternal health conditions in the country.

Previous research has linked short birth intervals to factors such as age [13], maternal education [13-15], residence [2], wealth index [2, 14], sex of the index child [14], perinatal death [15], breastfeeding [2, 14], contraception [2, 14] and parity [13]. However, complex factors such as socioeconomic, demographic, political, cultural, population, and health services have been noted to have an impact on IPI [16]. To the best of our knowledge, despite access to services, the factors contributing to short IPI in urban settings are unclear.

This study was proposed to fill this knowledge gap by examining the effect of clustering using a shared frailty modeling approach in five geographically diverse urban settings. Therefore, this study was aimed to assess the duration of the inter-pregnancy interval and its predictors among pregnant women in urban South Ethiopia. The findings will help to strengthen interventions aimed at optimizing inter-pregnancy intervals, such as family planning programs. It will also help to accelerate maternal and child health-related sustainable development goals.

Methods

Study setting

This research was carried out in the Hadiya zone, which is located in the Southern Nations, Nationalities and Peoples Region (SNNPR), Ethiopia. The administrative center of the zone is Hossana town. In Hadiya zone, there is one general hospital, three primary hospitals, 62 health centers, and 311 health posts that provide health services to the community [Hadiya Zone Health Bureau report-Unpublished]. In this study, five urban settings (Hossana, Shone, Gimbichu, Jajura, and Homecho), which consist of a total of eighteen kebeles, were included. Kebele is the lowest administrative unit in Ethiopia. It is a part of a district or sub-district that contains households or a delimited group of people.

Study design and population

A community-based retrospective follow-up study was carried out in five urban settings in Hadiya zone, South Ethiopia. The study was conducted among pregnant women who had a live birth during their most recent childbirth from July 1, 2014 onwards (i.e., women who had their most recent live birth within the last five years preceding the start date (July 8, 2019) of data collection). All women who met the following eligibility criteria were included: had at least one live birth during the most recent childbirth; had no recent abortion; had no recent stillbirth; could recall the date of recent childbirth or could show an immunization card; and were pregnant at the time of data collection.

Sample size and data collection techniques

A sample size of 440 was calculated in Epi Info StatCalc version 7.2.2.6 software using the formula for cohort or cross sectional designs, assuming % outcome (short IPI) in the unexposed group (contraceptive non-users) = 65.1%, % outcome (short IPI) in the exposed group (contraceptive users) = 51.5% from the previous study [2], two-sided confidence level = 95%, and power of 80%. We calculated IPI by subtracting nine months of gestational age from the birth interval [1]. However, from July 8, 2019 to December 30, 2019, a total of 2171 pregnant women were identified through house-to-house identification and included in the study. Face-to-face interviews conducted at house-hold level during the identification, using a structured questionnaire.

Variables and measurements

The outcome variable was the duration of the inter-pregnancy interval. The inter-pregnancy interval was calculated by subtracting the date of the most recent childbirth from the date of the last menstrual period (IPI in months = date of LMP minus date of recent childbirth). When women were unable to recall the date of their last menstrual period, gestational age was estimated using Ultrasound and then subtracted from the date that a woman had the Ultrasound scan to obtain the date of LMP (date of LMP = date of Ultrasound scan minus the gestational age at the time of the Ultrasound scan) [1]. A follow-up period was defined as the number of months between a live birth and conception or the women’s LMP. The event (failure) was defined as the occurrence of pregnancy after a live birth within 24 months (short IPI), whereas the censored (success) was the absence of pregnancy within 24 months (optimal IPI) [1]. The duration of months spent from live birth to subsequent pregnancy, or the woman’s LMP, was a time variable. The independent variables included in this study were: 1) socio-demographic and economic, such as religion, ethnicity, marital status, maternal age, sex of child, the number of children, education, occupation, and wealth index. 2) Reproductive characteristics such as parity, age at first childbirth, mode of delivery, and breastfeeding 3) Health-related services such as contraceptive use, antenatal and postnatal care visits, counselling, place of delivery. 4) Decision-making such as decision-making for contraceptive use, discussion with husband about pregnancy spacing and whether the husband encourages wife to space pregnancies or not (Table 1).

Table 1

List of variables, definitions and measurements for the study in urban South Ethiopia, 2019.

Variables	Measurements
Inter-pregnancy interval (IPI)	Time duration from date of live birth to date of woman’s last menstrual period in months. It was categorized as event if pregnancy occurred in <24 months (short IPI) or censored if it occurred at ≥24 months (optimal IPI). The categorization was based on World Health Organization recommendation for pregnancy spacing [1].
Maternal age	The reported age of a woman at the time of the interview in completed year. It was categorized as 20–24, 25–29 and ≥30 years
Maternal education status	Education level of a woman at the time of interview. It was categorized as no formal, primary, secondary and higher education
Age at first childbirth	Reported age at the time that a woman had her first childbirth. It was categorized as <20 and ≥20 years.
Greater number of children by sex	Whether the family has more female, male or equal number of children by sex, which was categorized as female, male and equal
Parity	The number of times a woman has given birth, regardless of the outcomes. It was classified as 1, 2, 3 and ≥4
Number of previous ANC visits	The number of antenatal care visits a woman made during her previous pregnancy. It was categorized as <4 and ≥4.
Recent number of children	The total number of children that the family recently has, categorized as 0–1, 2–3 and ≥4.
Past history of Stillbirth	Whether a woman had a history of giving birth to a baby with no signs of life such as no breathing, no heartbeat, and no movement prior to the most recent delivery. It was answered as ’yes’ if present, ’no’ otherwise.
Survival status of the recent child	Whether the most recent child was alive or died, responded as ‘yes’ if alive, ‘no’ if died.
Counselled during previous ANC visits	Whether or not the woman received advice from health care providers about spacing pregnancies during previous antenatal visits, responded by a yes or no response.
Counselled during PNC	Whether or not the woman received advice from health care providers about spacing pregnancies during the postnatal period, including child immunization visits, was responded by a yes or no response.
Exclusive breastfeeding	If the mother had only breast milk for her most recent child for up to six months, without any additional food or fluids, except medications, the answer was yes if <6 months and no if ≥6 months.
Total duration of breastfeeding	Women were asked how long they had breastfed their most recent child until it was discontinued in months, and then the reported number of months categorized as <24 and ≥24.
Decision maker on contraception	Whether a wife, husband, or both made the decision to use contraception when it was necessary, it was reported as wife alone, husband alone, and jointly (both).
Discussion with husband	If the wife has discussed or talked with her husband about spacing pregnancies after a recent childbirth, responded as yes or no.
Husband encourages spacing	Whether a husband encourages his wife to space pregnancies via safe methods of contraception when she requested and/or himself advise her to use responsibly, responded as yes or no.
Modern contraceptive use	If a woman used any form of modern contraception after recent childbirth, responded as yes or no
Plan to wait until current pregnancy	Refers to how long couples planned (if they planned) to wait between their most recent childbirth and their current pregnancy in months, categorized as <24 and ≥24
Desired number of children	Refers to the number of children that the couples (both husband and wife) wish to have in agreement, categorized as 1–5, ≥ 6 and undecided. If the woman and her husband could not agree on the number of children or had different wishes, it was classified as "undecided."
Mode of delivery for the recent child	Refers to the process of delivery for the most recent child, and categorized as spontaneous vaginal, cesarean section and instrumental (forceps or vacuum) delivery.

The wealth index was measured using household assets for urban residences, which include the following items: the owner of the house; the number of rooms; the material of the roof; the material of the floor; the material of the exterior wall; the source of drinking water; the type of latrine; the type of cooking materials (1 = electricity, 0 = wood/charcoal/biogas/natural gas, etc.), the source of income, and the presence or absence of: cell phone, refrigerator, radio, television, stove, chair, table, watch, modern bed, bicycle, Bajaj (three-wheeled vehicle), motor cycle, car, donkey/horse cart, and bank account. Each item was categorized into two (1 = yes and 0 = no). Based on the World Food Program and WHO recommendations, latrines and water sources were categorized as improved and unimproved facilities [17]. Principal component analysis was done to generate the components. Finally, ranking was done in five categories (lowest, second, middle, fourth, and highest).

Quality control measures

The questionnaire was developed in English from related literature and the Ethiopia Demographic and Health Survey (EDHS) and translated into the Amharic language. It was pre-tested in a similar setting (Durame Town). Two days of training were given for data collectors (ten midwives) and supervisors (five public health professionals) on the concept and approaches to the participants. Supervisors closely monitored the data collection process. To minimize recall bias related to recalling the date of the last childbirth, we limited the date of the most recent childbirth to the previous five years. Family members such as the husband, grandparents, and mother-in-law were also involved to recall the date of childbirth. To reduce selection bias, all pregnant women during the study period were included based on predetermined eligibility criteria. Furthermore, Ultrasound was used for women who had difficulty remembering the date of their LMP for a variety of reasons, including contraceptive use and breastfeeding. Epi-data was used to control data entry errors.

Analysis

The data were entered into Epi-data version 3.1 and analyzed in Stata version 14. Prior to analysis, the data were explored to check outliers and missing values. For continuous variables, descriptive statistics such as mean, median, and standard deviations were calculated. For categorical variables, frequencies and percentages were computed. A complete case analysis was performed for the missing data. A survival analysis model was fitted since IPI is a time to event variable (from live birth to pregnancy). The Kaplan-Meir or product limit estimator was used to estimate cumulative survival probabilities and compare survival for the predictors. A Log-rank test was used to test the quality of survival between different groups and see if the graphs were significantly different for predictors of IPI. Because the data came from 18 different clusters (kebeles), the clustering effect (between cluster variations) was examined using the frailty variance of theta in the Cox gamma shared frailty model (null model). Kebele was used as a clustering variable.

Variables that showed a statistically significant association with short IPI at P<0.20 in the bivariable Cox gamma shared frailty model were selected for adjustment in the multivariable model. Variables that showed a statistically significant association at P<0.05 and 95% CI for adjusted hazard ratio that did not include 1 in the multivariable Cox gamma shared frailty model were reported as predictors of short IPI. In the adjusted model, interaction for possible effect modification was checked. A model with a better fit was selected by using log-likelihood, Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC). The results were interpreted using the hazard ratio (HR) as an effect measure. A clustering effect was estimated using the median hazard ratio (MHR).

Ethics approval and consent to participate

Ethical approval was obtained from the Institutional Review Board (IRB) of the University of Gondar, with registration number: O/V/P/RCS/05/1051/2019. Permission was obtained from regional and local health offices. The study participants were informed about how they were included in the study, the purpose of the study, their rights to withdraw or continue, and the potential benefits and harms of the study. A written consent form was prepared and attached together with the questionnaire to obtain approval from each study participant by signature or fingerprint.

Results

Socio-demographic and reproductive characteristics

A total of 2202 pregnant women who fulfill the inclusion criteria in the five urban settings were identified. Of these, 31 women were refused to participate. This corresponds to a response rate of 98.6%. The study included the remaining 2171 pregnant women. The gestational age of the pregnant women at the time of the interview ranges from 12 to 24 weeks. The majority, 2149 (99%), of the pregnant women were in the second trimester of pregnancy. The rest, 22 (1%), were at the end of the first trimester of pregnancy. The mean age of the pregnant women was 27.31 ± 3.44 years. The ages range from 20 to 40 years. The majority, 1142 (54.5%), belong to the age group of 25–29 years. Among the participants, 1936 (89.3%) were Hadiya in ethnicity (.

Duration of inter-pregnancy interval

The median duration of IPI was 22, 95% CI (21, 23) months, with an inter-quartile range of 14 months. During the follow-up, a total of 1199 (55.2%) pregnancies occurred within a short interval (<24months). The cumulative probability of survival (remaining not pregnant) decreased as the months after live birth increased (

Predictors of short inter-pregnancy interval

For these clustered data, shared frailty or unobserved heterogeneity due to clustering was checked by fitting a Cox gamma shared frailty model (null model). The Cox gamma shared frailty model (null model) with efron (method to handle tied failures) produced a higher Log-likelihood (-8772.261), lower AIC (17544.52) and BIC (17544.52), and was selected as a good fit model for our data. In the Cox gamma shared frailty model, the null model frailty variance of theta equals 0.05, 95% CI (0.01, 0.09), LR test of theta = 0: X2 = 44.15, P<0.001 indicates that there is unobserved heterogeneity or shared frailty, as frailty variance of theta and its 95% CI are greater than zero. That means, women from the same cluster are more or less prone to have short IPI, and assuming dependency (correlation) within a cluster and variation between clusters would yield more reliable estimates of the predictors. In the bivariable Cox gamma shared frailty model, maternal age, maternal education, parity, age at first childbirth, greater number of children by sex, recent number of children, desired number of children, number of previous ANC visits, past history of stillbirth, survival status of recent child, counseling during ANC, counseling during PNC, exclusive breastfeeding, total duration of breastfeeding, mode of delivery for the recent child, decision-making for contraception, discussion with husband, husband encouragement for pregnancy spacing, modern contraceptive use, planning pregnancy, and wealth status were significantly associated with short IPI at P<0.20.

In multivariable Cox gamma shared frailty model, maternal age ≥30 years, having no formal education, contraceptive non-use, short duration of breastfeeding, death of recent child, planning pregnancy for <24 months, not discussing with husband about pregnancy spacing and husband not encouraging pregnancy spacing were found to be statistically significant predictors of short IPI, with a 95% confidence level and P<0.05.

Accordingly, women who were 30 years or older were 25% [AHR = 0.75, 95% CI: 0.58, 0.97] less likely to have a pregnancy within a short period of time after giving birth than women who were 20–24 years old. Women with no formal education were 40% [AHR = 0.60, 95% CI: 0.46, 0.78] less likely to have a pregnancy within a short period of time after giving birth than those with a higher education. Women who did not use any modern contraceptive methods after a recent childbirth were twice [AHR = 2.27, 95% CI: 1.94, 2.66] more likely to become pregnant shortly after childbirth than contraceptive users. Women who stopped breastfeeding their most recent child within 24 months were nearly five times more likely to become pregnant in a short period of time [AHR = 4.92, 95% CI: 3.95, 6.12] than those who continued breastfeeding for the recommended duration (>24 months). Women who had lost the most recent child (in the first month of life) were nearly three times [AHR = 2.90, 95% CI: 1.41, 5.97] more likely to have a pregnancy within a short period of time than women whose most recent child was alive. Women who planned to become pregnant within 24 months after a live birth were nearly twice [AHR = 1.72, 95% CI: 1.26, 2.35] more likely to have a pregnancy within a short duration than women who planned for 24 or more months. Women who had no discussion with their husband about pregnancy spacing were 33% [AHR = 1.33, 95% CI: 1.10, 1.60] more likely to have a pregnancy within a short duration after childbirth than their counterparts. Similarly, women whose husband did not encourage pregnancy spacing were 25% [AHR = 1.25, 95% CI: 1.05, 1.48] more likely to have a pregnancy within a short duration after live birth than women whose husband encouraged pregnancy spacing.

Adjusting for predictors, the median increase in the hazard of short IPI when comparing a woman in a cluster (kebele) with a higher short IPI to a woman in a cluster with a lower short IPI was 30% [MHR = 1.30, 95% CI: 1.11, 1.43] [18] (.

Discussion

In this study, the median duration of IPI in urban settings was found to be short. Maternal age ≥30 years and having no formal education were protective factors of short IPI. Non-use of modern contraceptive methods, short duration of breastfeeding, death of the recent child, planning pregnancy within a short duration, not having discussion with the husband and husband not encouraging pregnancy spacing were found to increase the chance of having a pregnancy within a short duration after birth.

Despite the attempts made to minimize it, this study might have limitations. This study relied on retrospective follow-up data, so it might have had recall bias. Estimating gestational age using ultrasound and LMP might give different intervals to some extent. Hence, the cumulative survival graph and survival probabilities need to be interpreted taking this into consideration. Despite the limitations, it will provide useful information for planning maternal health services.

The estimated median duration (22 months) of IPI in urban settings was lower than the estimates of Lemo, 24 months [2] and Dabat, 23.6 months [19] districts and the national demographic and health survey, 25.5 months [4], Tanzania, 24.4 months [13] and Bangladesh, 46 months [15]. The variations could be due to differences in estimating IPI using LMP. We calculated IPI based on the date of LMP and the Ultrasound results of pregnant women. For this discussion, IPI from those studies was estimated by subtracting nine months of gestational age from the birth interval. Population characteristics, study settings, and sample size could be the other reasons. The desire for the number of children might be varied in different settings, even within the same country. Individual preferences for rearing children might vary; some people might wish to have children too closely so that they reach the desired number of children within a few years and then go on to their business, such as education and income-generating activities to support their family. The estimated duration of IPI in this study was less than the WHO recommended duration of 24 months [1]. This indicates that the presence of access to health services in urban areas cannot assure the use of modern contraceptive methods to space pregnancies adequately.

In this study, women who were 30 years or older were less likely to have a pregnancy within a short duration after childbirth as compared to those belong to 20–24 years of age. As the ages of women increased, possibly they might have the desired number of children or family size than those belong to an earlier-ages. It could also be due to fertility decline or delays; as women age increases (>30 years), fertility is likely decreased or delayed [20]. On the other hand, the younger age group, especially those who had lower birth order, would have limited experiences of fertility regulation, fewer exposures to health facility visits for maternal health services, and subsequent counseling on the need to space pregnancies. Thus, they might have fewer experiences with pregnancies and childbirth difficulties as well [15, 21]. This finding is consistent with the findings of other studies, which found that older women were less likely to have shorter intervals [13, 22].

Women with no formal education were less likely to have a pregnancy within a short period of time after birth than women with a higher level of education. Previous studies in Tanzania [13], Ethiopia [14], and Iran [21] found that less educated women had shorter birth intervals. However, the finding is consistent with that of Bangladesh [15] and Korea [23], which found that women with a higher level of education were more likely to have a short birth interval. Higher educated women might have a better employment opportunities, hence might breastfeed less frequently due to lower contact time or lack of breaks, far away work places, full-time employment, and inflexible working time that contribute to the fast return of fertility and increased risk of getting pregnant [24-26]. On the other hand, they might have further education plans so that they might stick to the schedule and wish to have the desired number of children in a few years and give other breast milk substitutes instead of breastfeeding for a longer duration [27].

Women who did not use modern contraceptive methods after a recent childbirth were more likely to become pregnant within a short period of time than contraceptive users, which is consistent with other studies [2, 19, 21, 28, 29]. Contraception is the main tool to achieve optimal pregnancy interval, but can be affected by various factors that health service programs have to address, including myths and misconceptions (mainly fear of infertility) [30].

Women who breastfed their most recent child for a shorter period of time (24 months) had a pregnancy sooner than those who breastfed for the optimal period of time (≤24 months). The finding is consistent with previous research in South Ethiopia [2], Bangladesh [15], Iran [21], and North Ethiopia [28]. This suggests breastfeeding has a positive impact on the duration of IPI, and encouraging women to prolong breastfeeding duration is beneficial. When a woman exclusively breastfeeds her child, it can delay the return of menses and promote pregnancy spacing [31, 32]. Extending breastfeeding after exclusive breastfeeding helps to space pregnancies while also increasing child survival [33]. Although personal, cultural, social, and environmental factors can influence breastfeeding frequency and duration [34], Ethiopia has a good culture of breastfeeding for a longer duration that needs to be promoted for further achievements in increasing pregnancy interval [35].

Modern contraception and exclusive breastfeeding have a positive (agonistic) interaction. Women who did not use modern contraception methods and did not exclusively breastfeed had a higher risk of becoming pregnant soon after birth [AHR = 1.82, 95% CI: 1.17, 2.82]. This indicates that modern contraceptive use and exclusive breastfeeding are highly effective strategies for spacing pregnancies.

Women who planned to become pregnant within 24 months after their recent childbirth were more likely to have a pregnancy within a short duration than those who planned for 24 or more months. This suggests that simply having a plan may not be sufficient; rather, how long is important in planning the next pregnancy. Those who had planned their next pregnancy for a longer period of time might have had their own reasons. As a result, they might have used contraception more consistently to achieve their goals than those who planned for a shorter period of time [29].

Women who lost their recent child (within the first months of life in this study) were more likely to have a short duration of inter-pregnancy interval than their counterparts. Women who lost their last child won’t breastfeed to benefit from lactation amenorrhea. Thus, they were more likely to have menses return soon and more likely to get pregnant unless contraceptive methods were used. Those women might also be encouraged by significant others to replace the lost child within a short duration of time, which might help them to recover from psychosocial impacts and forget the lost child. The finding was supported by the study conducted in North-West Ethiopia [19], in which the death of the index child contributed to the occurrence of births within a short period of time.

There was a positive interaction between planning pregnancy within a short duration after a live birth and the death of the index child [AHR = 5.30, 95% CI: 1.06, 26.54]. Women who had already planned to become pregnant soon after childbirth and, if their child died, might have a strong desire to replace the lost child soon.

Women who did not discuss about pregnancy spacing following recent childbirth had a pregnancy within a short duration compared to their counterparts. It is obvious that pregnancy and childbirth need a joint decision of couples/relatives. Women who had a discussion with their husband might have used postpartum contraception in order to space pregnancy [36, 37].

Women whose husbands did not encourage pregnancy spacing were more likely to have a pregnancy in a short period of time than women whose husbands did encourage pregnancy spacing. This suggests that the decision to space the pregnancy is made by the husband. Without the consent of her husband, a woman might not use any form of contraception to space pregnancies [37]. In Ethiopia, preference for the number of children might vary between couples, and the husband might have a greater wish for additional children than the woman [38]. It is also common for husbands to make decisions about most household issues and maternal health services. A husband’s involvement may be an important component of maternal health services, including family planning methods [29, 38].

The median hazard ratio indicated that there is variation in inter-pregnancy interval duration due to variation among clusters, highlighting the need for further investigation of contextual factors to implement evidence-based interventions in the urban community. Considering the limitations mentioned above, the findings of this study could be generalized to similar urban settings with similar populations.

Conclusions

Although maternal health services such as modern contraceptive methods and information are available in urban areas, this study found that more than half of women in urban areas have a shorter duration of IPI than the WHO recommendation. The short duration of IPI was associated with modifiable factors such as non-use of modern contraceptive methods, short duration of breastfeeding, planning pregnancy within a short duration after childbirth, and lack of decision-making autonomy of women to use maternal health services. Therefore, contraceptive utilization, duration of breastfeeding, and planning time of pregnancy have to be improved. Men’s involvement in reproductive health services and advocating reproductive decision-making autonomy of women are also fundamental, as decisions to use maternal health services like contraception are usually made by men. Women with higher education levels have to give emphasis to optimal pregnancy spacing. Taking into account contextual differences in inter-pregnancy interval when providing maternal and child health services may help to reduce the risk of short inter-pregnancy interval in the community.

Kaplan-Meir survival graphs for the predictors of short IPI.

(DOC)

Click here for additional data file.

Log-rank test for the potential predictors (unadjusted) of short IPI.

(DOCX)

Click here for additional data file.

R code to calculate median hazard ratio.

(DOCX)

Click here for additional data file.

Dataset.

(DTA)

Click here for additional data file.

13 Jan 2022

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Thomas Phillips, PhD

Staff Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please provide a sample size and power calculation in the Methods, or discuss the reasons for not performing one before study initiation.

3. Thank you for stating the following financial disclosure: "The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript."

At this time, please address the following queries:

a) Please clarify the sources of funding (financial or material support) for your study. List the grants or organizations that supported your study, including funding received from your institution.

b) State what role the funders took in the study. If the funders had no role in your study, please state: “The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

c) If any authors received a salary from any of your funders, please state which authors and which funders.

d) If you did not receive any funding for this study, please state: “The authors received no specific funding for this work.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

4. Thank you for stating the following in the Acknowledgments Section of your manuscript:

"We would like to thank University of Gondar and Wachemo University for financial support.  We are very much thankful for study participants, data collectors and supervisors for their contributions."

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

"The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript"

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

5. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially sensitive information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

6. PLOS requires an ORCID iD for the corresponding author in Editorial Manager on papers submitted after December 6th, 2016. Please ensure that you have an ORCID iD and that it is validated in Editorial Manager. To do this, go to ‘Update my Information’ (in the upper left-hand corner of the main menu), and click on the Fetch/Validate link next to the ORCID field. This will take you to the ORCID site and allow you to create a new iD or authenticate a pre-existing iD in Editorial Manager. Please see the following video for instructions on linking an ORCID iD to your Editorial Manager account: https://www.youtube.com/watch?v=_xcclfuvtxQ

7. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: I Don't Know

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: No

Reviewer #2: No

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The article presents original research. As far I could verify, the results are not published elsewhere. It addresses a significant problem of children, women and community as women in the described community are responsible for both indoor and outdoor responsibilities. Methods and materials are well described and are consistent with the type of the study. The authors used advanced analysis appropriate for data nature to obtain correct information from the data. However, before considering publication the paper will be benefited from the following comments.

1. Pleas add cluster effect analysis

2. study design, population and quality control (better to add this for title population since the paragraph contain these)

3. why you excluded those gave birth before 15 years. Your study population may be reproductive age women, however, you can interview these women but they may have births before 15 years. which may be the potential factor not to use contraceptive and at all to had no power to decide on their reproductive health.

4. starting from line 153 to paragraph before quality control is part of result but not analysis part.

5. table 2 is part of result but not the method part, so please move into result section.

6. first what is the importance of checking PH assumption as such correlated (have common shared value at cluster ) level. rather it is good to check whether frailty is significant or none significant. if frailty is not significant PH ascription test is ideal to fit cox , stratified cox or parametric model. if frailty is significant cox frailty model is ideal. second to consider parametric model what exploratory analysis do you have conducted to support your justified model (to see this please demonstrate base line hazard (h(to) ) distribution and other parametric assumptions. why you are limited on only (Exponential, Gompertz and Weibull), what about at least log normal and loglogistic, and generalised gamma since ststa support these also.

7. what variable selection method do use (authomethod method or purposive method, if it was purposive method 20 to 25% not 5% for variable selection or specify your reference) ???? 0.25 or 0.2 from bivariable to multivariable model. (or log rank test 0.05)

8. Quality control measures (it should come before analysis)

9. From line 191 to 193 (It is part of analysis but not quality control measure)

10. In your method section you have stated (line 171-173) Bivariable weibull inversgaussian shared frailty analysis before Multivariable analysis but here in the result section from line 219 to 228 you have stated log rank test as a selection of variables before multivariable analysis. so please make consistent and if you had used log rank test please express your method for continuous predictors.

11. you can put this (table 4) as supporting document

12. I have seen your interpretation in your document 1. do you think the interpretation of crude and adjusted hazard ratio interpretation similar? I think this interpretation will work for crude one not for adjusted HR. 2. you had also adjust for clusters but your interpretation did not show this and 3. I recommend to conduct median HR to conduct for the cluster effect.

13. in the method section you had mentioned that you consider effect modification but I haven't seen in result part and your result like Survival status of the recent child and Plan to wait until current pregnancy invite interaction analysis consideration since there is beg CHR and AHR discrepancy.

14. In all document please merge citations (that have more than one citations)

15. Line 296 to 298 as far as you have adjusted for Decision maker for contraception and discussion with husband, as well as wealth how do you consider these as confounder or possible reason.

16. Line 310 to 311 need reference

17. Line 313 to 314, is it study based? If it is please cite.

18. Line 318 to 319 what is your base to consider interaction (your statistical or biological or clinical mechanism), what about other variable interactions, since your CHR and AHR for Survival status of the recent child, Plan to wait until current pregnancy ... show more than 15% crude adjusted discrepancy to consider interaction or confounding.

Reviewer #2: GENERALCOMMENTS: There are several grammatical errors and wrong use of tenses hat need to be corrected in order to make the text more understandable.

METHODS: The a118-119 authors should explain what kebeles are for better understanding. What informed the choice of the 5 urban settings selected for the study? In page 6, lines who performed the ultrasound scans, where and when was this done? The gestational ages of the women should have been added for clarity. In page 6,lines 119=120 it is not clear how the interpregnancy interval was determined from the USS. What about those who were in advanced pregnancy when ultrasonography cannot date pregnancy accurately and did not remember their LMP. In page 9,line 140-the authors should explain what baja is.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Reta Dewau

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: PONE-D-21-12390_reviewer.pdf

Click here for additional data file.

10 Feb 2022

Editor’s comments:

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming.

Authors’ response: Dear editor, thank you again for your time and contribution. We looked at the PLOS ONE style requirements to edit the whole manuscript and revised accordingly.

2. Please provide a sample size and power calculation in the Methods, or discuss the reasons for not performing one before study initiation.

Authors’ response: Thank you dear editor, we included the sample size and power calculation in methods section.

3. Thank you for stating the following financial disclosure: "The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript."

At this time, please address the following queries:

a) Please clarify the sources of funding (financial or material support) for your study. List the grants or organizations that supported your study, including funding received from your institution.

b) State what role the funders took in the study. If the funders had no role in your study, please state: “The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

c) If any authors received a salary from any of your funders, please state which authors and which funders.

d) If you did not receive any funding for this study, please state: “The authors received no specific funding for this work.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

Authors’ response: Dear editor, we have revised the financial disclosure statements. We revised as “The authors received no specific funding for this work.”

4. Thank you for stating the following in the Acknowledgments Section of your manuscript:

"We would like to thank University of Gondar and Wachemo University for financial support. We are very much thankful for study participants, data collectors and supervisors for their contributions."

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

"The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript"

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

Authors’ response: Dear editor, we have removed funding-related text from the manuscript. We included this in the revised cover letter.

5. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially sensitive information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

Authors’ response: Dear editor, we have uploaded dataset as supporting information (S1 File. Dataset).

6. PLOS requires an ORCID iD for the corresponding author in Editorial Manager on papers submitted after December 6th, 2016. Please ensure that you have an ORCID iD and that it is validated in Editorial Manager. To do this, go to ‘Update my Information’ (in the upper left-hand corner of the main menu), and click on the Fetch/Validate link next to the ORCID field. This will take you to the ORCID site and allow you to create a new iD or authenticate a pre-existing iD in Editorial Manager. Please see the following video for instructions on linking an ORCID iD to your Editorial Manager account: https://www.youtube.com/watch?v=_xcclfuvtxQ

Authors’ response: Thank you dear editor, the corresponding author’s ORCID ID was created.

7. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

Authors’ response: Dear editor, thank you for giving useful link to edit the supporting information. The caption for supportive information files was added in the manuscript, and also inside each uploaded supportive documents. We also used Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool to revise figure inside the manuscript.

Reviewers’ comments

Reviewer #1

The article presents original research. As far I could verify, the results are not published elsewhere. It addresses a significant problem of children, women and community as women in the described community are responsible for both indoor and outdoor responsibilities. Methods and materials are well described and are consistent with the type of the study. The authors used advanced analysis appropriate for data nature to obtain correct information from the data. However, before considering publication the paper will be benefited from the following comments.

1. Pleas add cluster effect analysis

Authors’ response: Dear reviewer, thank you for suggesting us to include a newly introduced effect measure (the median hazard ratio) to understand the magnitude of clustering effect. Reporting cluster effect, rather than reporting only the variance and individual characteristics, is useful to understand the importance of cluster for evidence-based interventions.

Thus, we added cluster effect analysis according to your recommendation. We included this in methods section (in the analysis) and result section. Kindly, see page number 12 and 21-22.

2. study design, population and quality control (better to add this for title population since the paragraph contain these)

Authors’ response: Dear reviewer, we have revised this section as ‘study design and population’. The quality control contents were moved to quality control section.

3. why you excluded those gave birth before 15 years. Your study population may be reproductive age women, however, you can interview these women but they may have births before 15 years. which may be the potential factor not to use contraceptive and at all to had no power to decide on their reproductive health.

Authors’ response: Dear reviewer, thank you for the comment. We did not exclude those with age below 15 years. In our study, we included those women who had at least one live birth before current pregnancy, irrespective of their age at first childbirth. We just categorized ‘age at first childbirth’ (15-19 and ≥20 years) after data collection, for the analysis purpose. That means, we did not get a woman whose age at first childbirth was less than 15 years. We revised it as <20 and ≥20 years, this may give more sense.

4. starting from line 153 to paragraph before quality control is part of result but not analysis part.

Authors’ response: Thank you dear reviewer, we moved this to the result section.

5. table 2 is part of result but not the method part, so please move into result section.

Authors’ response: Thank you dear reviewer, we removed table 2 which was about Cox proportional hazard assumption. As you said below in comment number 6 “since frailty is significant, testing Cox proportional hazard assumption was not relevant”. Thus, in the revised manuscript we removed it.

6. first what is the importance of checking PH assumption as such correlated (have common shared value at cluster ) level. rather it is good to check whether frailty is significant or none significant. if frailty is not significant PH ascription test is ideal to fit cox , stratified cox or parametric model. if frailty is significant cox frailty model is ideal. second to consider parametric model what exploratory analysis do you have conducted to support your justified model (to see this please demonstrate base line hazard (h(to) ) distribution and other parametric assumptions. why you are limited on only (Exponential, Gompertz and Weibull), what about at least log normal and loglogistic, and generalised gamma since ststa support these also.

Authors’ response: Dear reviewer, thank you very much for suggesting us to follow the ideal approach for the survival analysis. We made significant revision in the analysis considering your constructive comment. Of course, in previous manuscript, our approach was parametric (Weibull inverse Gaussian shared frailty analysis). We choose parametric because Cox proportional assumption was violated. Then, we checked presence of frailty (with parametric approach), it was significant and Weibull inverse Gaussian shared frailty analysis was done. Weibull, Gompertz and exponential distributions were selected because they give hazard ratio. But other distributions (log normal, log logistic and generalized gamma) do not give hazard ratio. Rather they give rate ratio which makes the interpretation difficult. Since our aim is to report effect size using hazard ratio we compared the above three distributions and using Weibull found to be good fit model.

However, in this revised manuscript, we accepted your suggestion to consider a semi-parametric (Cox gamma shared frailty analysis) which is an ideal approach. As you said, in the revised manuscript we first checked whether the frailty is significant or not, using semi-parametric (Cox gamma) frailty. It was significant (LR test of frailty variance was significant (P<0.001)). The null model frailty variance itself was greater than zero (0.0462576, 95% CI (0.007703, 0.084812)). Thus, shared frailty analysis is appropriate. Testing Cox proportional hazard assumption was not relevant once frailty is significant. So we removed the table that represented cox proportional hazard assumption. Therefore, the analysis in the revised manuscript was according to your suggestion (Cox gamma shared frailty). Thank you again for sharing your experience.

7. what variable selection method do use (authomethod method or purposive method, if it was purposive method 20 to 25% not 5% for variable selection or specify your reference) ???? 0.25 or 0.2 from bivariable to multivariable model. (or log rank test 0.05)

Authors’ response: Thank you dear reviewer for the constructive comment. For mixed effect models, including frailty, it is customary to use P vales such as 0.20 and 0.25 to select variables from bivariable to multivariable model rather than just using 0.05, as this (0.05) cutoff results in loss of important variables that were confounded by others. After your useful comment, we revised the analysis and used P<0.20 as a cutoff to select variables from bivariable to multivariable model. This helps to include as many predictor variables as possible for the adjustment. Of course, it is also possible to use the log-rank test but it did not give a hazard ratio to understand the strength of association. However, we used log-rank test to test the equality of survival across the different categories of a predictor variable. This was reported in supporting information file (S1 Table).

8. Quality control measures (it should come before analysis)

Authors’ response: Thank you dear reviewer we moved it before analysis section.

9. From line 191 to 193 (It is part of analysis but not quality control measure)

Authors’ response: Thank you dear reviewer we moved it from the quality control measure to the analysis section.

10. In your method section you have stated (line 171-173) Bivariable weibull inversgaussian shared frailty analysis before Multivariable analysis but here in the result section from line 219 to 228 you have stated log rank test as a selection of variables before multivariable analysis. so please make consistent and if you had used log rank test please express your method for continuous predictors.

Authors’ response: Thank you dear reviewer we revised this inside manuscript. We used bivariable Cox gamma shared frailty model to select variables for multivariable model, using P<0.20, as we responded for comment number 6 and 7 above. Now, in the revised manuscript, log-rank test was used only to test equality of survival across different categories of a variable, as reported in supporting information (S1 Table). Continuous variables were categorized or recoded before log-rank test.

11. you can put this (table 4) as supporting document

Authors’ response: Dear reviewer we put it as a supporting information (S1 Table).

12. I have seen your interpretation in your document 1. do you think the interpretation of crude and adjusted hazard ratio interpretation similar? I think this interpretation will work for crude one not for adjusted HR. 2. you had also adjust for clusters but your interpretation did not show this and 3. I recommend to conduct median HR to conduct for the cluster effect.

Authors’ response: the interpretation of hazard ratio in crude and adjusted model is similar, except the difference in the estimation. In the adjusted model, the hazard ratio is adjusted for other covariates. So only the estimates vary in the crude and adjusted model. The interpretation is similar with other effect measure, like relative risk. Both in crude and adjusted models, HR>1 indicate increased risk; HR<1 indicates decreased risk or protective effect of exposure, and HR=1 indicates no difference or no effect. Of course, interpretation of hazard ratio in parametric model is different (as it take accelerated failure time form) from a semi-parametric one. We tried to make the interpretation of hazard ratio easier to understand, especially for those readers outside the field. For example, the interpretation “the hazard of short inter-pregnancy interval was 2 time higher for women who did not use modern contraceptive methods than contraceptive users” may be difficult (‘the hazard of’) to understand by readers outside the field. This can be simplified as “Women who did not use modern contraceptive methods were 2 times more likely to become pregnant within short duration after childbirth than contraceptive users.

Regarding adjustment for cluster, we were just to say considering for clustering in univariable and multivariable models. We removed the phrase ‘adjusting for clustering’ from the text, it may not be important to report once we consider cluster effect analysis.

We conducted a median hazard ratio for the cluster effect. In this analysis, we reported both null model and full (adjusted) model median hazard ratios with their interpretations (kindly, see page number 21-22). Thank you again for encouraging us to consider this new effect measure for clustered data.

13. in the method section you had mentioned that you consider effect modification but I haven't seen in result part and your result like Survival status of the recent child and Plan to wait until current pregnancy invite interaction analysis consideration since there is beg CHR and AHR discrepancy.

Authors’ response: Thank you dear reviewer for the constructive comment. Now we considered those variables for the interaction effect (Table 3). Also, we checked the other variables as well.

14. In all document please merge citations (that have more than one citations)

Authors’ response: Thank you dear reviewer now we merged the citations.

15. Line 296 to 298 as far as you have adjusted for Decision maker for contraception and discussion with husband, as well as wealth how do you consider these as confounder or possible reason.

Authors’ response: Thank you dear reviewer for the input. Now we revised in the text. As you said, we already adjusted for those variables so no need to use them as possible reason. Thus, removed from the text.

16. Line 310 to 311 need reference

Authors’ response: Dear reviewer now we added the reference.

17. Line 313 to 314, is it study based? If it is please cite.

Authors’ response: Dear reviewer now we included the citation.

18. Line 318 to 319 what is your base to consider interaction (your statistical or biological or clinical mechanism), what about other variable interactions, since your CHR and AHR for Survival status of the recent child, Plan to wait until current pregnancy ... show more than 15% crude adjusted discrepancy to consider interaction or confounding.

Authors’ response: Dear reviewer, thank for suggesting us to check other variables for possible interactions. Our base to consider the interaction was our theoretical knowledge. For example, breast feeding has a potential to delay ovulation due to the effect of prolactin hormone. Delay in ovulation reduce the chance of getting pregnancy. Hormonal contraceptive methods also have similar mechanism of action (delay ovulation), and then prevent occurrence of pregnancy. These hormonal contraceptive methods are a synthetic hormones which are similar with female hormones (progestin and estrogen). Therefore, breast feeding and at the same time using hormonal contraceptive methods might have synergetic effect (effect modification). Of course, effect modification is not a bias. Reporting the independent effect (by avoiding interaction effect) is useful to identify the independent effect of an exposure variables on the outcome. The other is statistical interaction which might occur due to the nature of data or the effect measure used. It may not necessarily have clinical importance (plausibility) but may affect the parameter estimates and bias the relationship between exposure and outcome, as the value of one covariate depends on the value of the other covariate. Thus, we checked for statistical interaction by fitting all covariates in the adjusted model one by one. If the interaction is significant, we retain the interacted variables in the model. If not significant, we did not consider the variables in the model. In this regard, the only variables found to be significant for the interaction were ‘exclusive breast feeding with contraceptive use’ (biological or clinical interaction), and ‘planning pregnancy within short duration with the survival status (death) of the last child’ (statistical interaction). Interaction between ‘planning pregnancy within short duration and death of last child’ could be statistical because it has no biological or clinical mechanisms rather it might be due to the fact that those women who had a plan to give next child soon after the preceding birth might be provoked by the death of the preceding child.

In addition to theoretical knowledge, as you said, the gap between crude and adjusted estimates could be a base to invite interaction or confounding.

Again, thank you for this vital contribution to improve the quality of the analysis.

Reviewer #2

GENERALCOMMENTS: There are several grammatical errors and wrong use of tenses that need to be corrected in order to make the text more understandable.

Authors’ response: Thank you dear reviewer we tried to correct grammatical error and use of tenses as much as possible in the revised manuscript.

Reviewer comment: METHODS: The a118-119 authors should explain what kebeles are for better understanding.

Authors’ response: Dear reviewer now we explained what the kebele is in the methods section (under study setting on page 5). Kebele is the lowest administrative unit in Ethiopia. It is just a part of district or sub-district. The administration system in Ethiopia is as follow: Federal�  Region�  Zone�  District�  Kebele. Next to kebele is the household or families/person.

Reviewer comment: What informed the choice of the 5 urban settings selected for the study?

Authors’ response: Dear reviewer, the 5 urban settings were selected for the study for the following reasons. Firstly, as we have mentioned in the ‘introduction’ there is scarcity of evidence about factors contributing to short inter-pregnancy interval in the urban areas even though access for maternal health services are available. Secondly, the 5 urban areas are the only ‘town administrations’ in the zone. Thus, we included all of them. Thirdly, in these urban areas there are hospitals with Ultrasound services. For those women who unable to recall date of last menstrual period, we use this opportunity to estimate gestational age and subsequently calculate date of last menstrual period and inter-pregnancy interval rather than excluding those who unable to recall the dates of last menstrual period. This helps to reduce selection bias related to recalling date of last menstrual period. Fourthly, others are semi-urban which are not that much different from the rural and no such ultrasound facilities to estimate gestational ages as many women in rural areas has difficulty in recalling the dates of last menstrual period.

Reviewer comment: In page 6, lines who performed the ultrasound scans, where and when was this done?

Authors’ response: Dear reviewer, thank you for the constructive comments. In our setting, it is not feasible to us to scan every woman by Ultrasound. Thus, the last menstrual period recall is mainly used. In this study, the Ultrasound scan was done at the hospitals of each town administration for free of cost (after communication with hospital administration Ultrasound service was allowed for those who could not be able to recall the date of last menstrual period). The Ultrasound scan was performed by those who already give routine Ultrasound scan in the hospitals (radiologists and trained medical Doctors). The Ultrasound scan was done (for those unable to recall LMP) at the end of first trimester and just at second trimester of pregnancy. We included pregnant women who were at the end of first trimester and at second trimester. We did so for two reasons; firstly, it is costly to us to test pregnancy in the first trimester (laboratory-based test). In first trimester, some women even may not know whether they are pregnant. Thus, we included women whose pregnancy was already confirmed. Secondly, women in the advanced pregnancy (third trimester) might have difficulty in recalling the date of last menstrual period, and the Ultrasound is less accurate. Hence pregnant women at third trimester were not included.

Reviewer comment: The gestational ages of the women should have been added for clarity.

Authors’ response: Dear reviewer, we included the gestational age of women (gestational age at the time of interview in weeks) in the revised manuscript. Kindly see the text on page 13.

Reviewer comment: In page 6,lines 119=120 it is not clear how the interpregnancy interval was determined from the USS.

Authors’ response: Thank you dear reviewer now we clarified it by including the formula that the inter-pregnancy interval was calculated.

To calculate inter-pregnancy interval, we need to have two time points; date of last childbirth and date of last menstrual period. If we have these two dates, we can simply subtract date of last childbirth from the date of last menstrual period (i.e. inter-pregnancy interval = date of last menstrual period minus date of last childbirth). This gives inter-pregnancy interval in complete months. If date of last menstrual period is not known, we can calculate it from the gestational age which is determined by ultrasound. This can be done by subtracting the gestational age from the date that a woman was seen by the ultrasound; i.e. date of last menstrual period=date of the ultrasound scan minus the gestational age at the time of Ultrasound scan. This clarification remarks were reported in the revised manuscript. Kindly, see page number 6.

Reviewer comment: What about those who were in advanced pregnancy when ultrasonography cannot date pregnancy accurately and did not remember their LMP.

Authors’ response: Dear reviewer, thank you for the comment. As we responded in the above comment, the Ultrasound scan was done at the end of first trimester and at second trimester of pregnancy since we included women in these gestational ages.

Reviewer comment: In page 9,line 140-the authors should explain what baja is.

Authors’ response: Thank you dear reviewer we explained it in the text; Bajaj is a ‘three wheeled vehicle’.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

7 Jun 2022

Submitted filename: Dear Authors.docx

Click here for additional data file.

9 Jun 2022

To: PLOS ONE Journal, Editorial Office

Subject: Submitting a revised version of manuscript and response to Reviewers.

Ref: Submission ID PONE-D-21-12390R1

Title of Article: " Duration of inter-pregnancy interval and its predictors among pregnant women in urban South Ethiopia: Cox gamma shared frailty modeling"

Authors:

Belayneh Hamdela Jena, Gashaw Andargie Biks, Yigzaw Kebede Gete, Kassahun Alemu Gelaye

We would like to thank the Editor for facilitating and giving the opportunity to revise our manuscript. We are also grateful to reviewers for sharing their views and constructive comments. The comments are very important which will improve the quality of our manuscript. The point-by-point responses for each of the comments and the revised manuscript are provided in the attached documents.

Regards,

The authors!

Editor’s comments:

Journal Requirements:

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Authors’ response: Thank you dear Editor for your time and contribution. We checked the references and no retracted references found.

Reviewers’ comments

Reviewer #1

Some comments also

1. The finding of cluster effect needs to be added in the multivariable Cox gamma shared frailty model result table, in the discussion, abstract and conclusion as it has policy implication.

Authors’ response: Thank you dear reviewer for your contribution. We acknowledge your comments. In the revised manuscript we added the cluster effect in all sections according to your suggestion (kindly see the highlighted text in the table, discussion, abstract and conclusion sections of the revised manuscript).

2. In the predictors section theta value better not more than three digits.

Authors’ response: Thank you dear reviewer now we have corrected the digits into three

3. Simply base on full model variance of theta and its 95 CI calculate the median hazard with 95%CI and put it at the end of multivariable analysis and interpret it in short, then better to remove the cluster effect analysis section after table 4.

(MHR=1.30 (95CI%,1.11-1.43))

Authors’ response: Thank you dear reviewer we have calculated the MHR and its 95% CI from full model variance of theta, and put the interpretation in short, under the multivariable analysis. We have deleted the cluster effect analysis section as well.

Reviewer #2:

GENERAL: There are still several grammatical errors and wrong use of tenses that need to be corrected throughout the text.

Authors’ response: Thank you dear reviewer for your contribution. We acknowledge your comments. In the revised manuscript we have attempted to correct those errors throughout the text as much as possible.

ABSTRACT: The last sentence in the results sub-section is rather confusing and should be recasted.

Authors’ response: Thank you dear reviewer now we have corrected it. The MHR is normally interpreted in that manner.

ITRODUCTION: Page 4, lines 1-2,it is not clear what the authors mean by pregnancies that were too long or short. Are they referring to duration of pregnancy or inter-pregnancy interval? In page 4, paragraph 2,line 1, it is not clear which recommendation the authors are referring to and the recommendation should be referenced.

Authors’ response: Thank you dear reviewer for suggesting to correct those vague words. In the revised manuscript we have corrected them. Too long and too short were referring to inter-pregnancy intervals. We have also clarified which recommendation we were referring, and cited as well. It refers world health organization recommendation of at least 24 months.

METHODS: Under sample size and data collection- line 4-5, the calculation of inter-pregnancy interval seams to be different from that in lines 1-3 under variables and measurements. When and where were the face to face interviews conducted and how were the identified houses selected?

Authors’ response: Thank you dear reviewer for the comments. We approximated inter-pregnancy interval from birth interval estimate, by subtracting 9 months of average gestational age, which is possible when information for inter-pregnancy interval is lacking, just to estimate sample size. For instance, birth interval 33 months is approximately equivalent to inter-pregnancy interval 24 months (33-9 = 24). Birth interval is could be estimated by subtracting the date of preceding childbirth (delivery) from the date of the recent childbirth (birth interval = date of recent childbirth minus date of preceding childbirth). Inter-pregnancy interval = date of last menstrual period minus date of the recent childbirth. The difference between birth interval and inter-pregnancy interval is the ‘gestational age’. Inter-pregnancy interval estimation does not include gestational age. So inter-pregnancy interval could be estimated from birth interval by subtracting the gestational age when information about inter-pregnancy interval is lacking. This was done because we lacked an inter-pregnancy interval estimates to calculate the sample size during design of the study. However, for our study we have both dates (date of last childbirth and date of last menstrual period or estimate from Ultrasound) information, and the population were pregnant women. The gestational age of women was also known so it is possible to estimate inter-pregnancy interval without birth interval information.

We have done a house-to-house visits in the urban community, and the face-to-face interviews were conducted during the visits, at the house-hold level. For this study, we included all eligible pregnant women who are willing to participate, during the study period. No sampling of houses was done, rather we considered all eligible pregnant women who meet predefined inclusion criteria.

DISCUSSION: Several grammatical errors. Page 23, paragraph 2, the last sentence should be recasted.

Authors’ response: Thank you dear reviewer we attempted to correct the grammatical errors in all sections of the manuscript. We preferred to remove the sentence as it is non-sensing to discuss in that manner.

END!

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

12 Jul 2022

Duration of inter-pregnancy interval and its predictors among pregnant women in urban South Ethiopia : Cox gamma shared frailty modeling

PONE-D-21-12390R2

Dear Dr. Jena,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Aniekan Abasiattai

Guest Editor

PLOS ONE

Additional Editor Comments (optional):

The authors have made the required corrections.

Reviewers' comments:

22 Jul 2022

PONE-D-21-12390R2

Duration of inter-pregnancy interval and its predictors among pregnant women in urban South Ethiopia: Cox gamma shared frailty modeling

Dear Dr. Jena:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Aniekan Abasiattai

Guest Editor

PLOS ONE

Table 2

Socio-demographic and reproductive characteristics of pregnant women in five urban settings in Hadiya zone, South Ethiopia, 2019.

Variables	Frequency (%)
Religion
Protestant	1898 (87.4)
Orthodox	115 (5.3)
Catholic	95 (4.4)
Muslim	52 (2.4)
Apostolic	11 (0.5)
Ethnicity
Hadiya	1936 (89.2)
Kembata/Tembaro	114 (5.3)
Guragie/Siltie/Amhara/Oromo	121 (5.5)
Marital status
Married	2133 (98.2)
Unmarried/ Divorced	38 (1.8)
Woman’s main occupation
Housewife	1613 (74.3)
Employed	339 (15.7)
Merchant/ farmer/daily laborer/waiter	219 (10)
Husband’s main occupation
Daily laborer	599 (27.6)
Merchant	554 (25.5)
Employed	663 (30.6)
Farmer	274 (12.6)
Driver	81 (3.7)
Husband’s education
No formal education	315 (14.6)
Primary school (1–8 grades)	830 (38.4)
Secondary school (9–12 grades)	487 (22.5)
Higher education	532 (24.5)
Gravidity
2–3	1462 (67.4)
≥ 4	709 (32.6)
Place of delivery for recent child
Home	369 (17)
Health facility	1802 (83)

Table 3

Multivariable Cox gamma shared frailty analysis for the predictors of short inter-pregnancy interval among pregnant women in urban South Ethiopia, 2019.

Variables	Pregnancy status		CHR (95% CI)	AHR (95% CI)
	Event (<24 months)	Censored (≥24 months)
Maternal age
20–24	218 (60.6)	142 (39.4)	1	1
25–29	633 (55.4)	509 (44.6)	0.88 (0.75, 1.03)▪	0.84 (0.69, 1.03)
≥ 30	307 (51.9)	285 (48.1)	0.77 (0.64, 0.93)**	0.75 (0.58, 0.97)*
Maternal education status
No formal education	209 (50.6)	204 (49.4)	0.94 (0.77, 1.16)	0.60 (0.46, 0.78)***
Primary	515 (55.9)	407 (44.1)	1.08 (0.91, 1.28)	0.82 (0.66, 1.00)
Secondary	278 (59.8)	187 (40.2)	1.21 (1.00, 1.45)*	1.01 (0.82, 1.25)
Higher	196 (53.1)	173 (46.9)	1	1
Age at first childbirth
< 20	283 (54.7)	234 (45.3)	1	1
≥ 20	890 (55.3)	718 (44.7)	1.21 (1.04, 1.39)*	1.18 (0.98, 1.42)
Greater number of children by sex
Female	531 (59.1)	365 (40.9)	1	1
Male	460 (54.1)	391 (45.9)	0.94 (0.83, 1.06)	1.03 (0.89, 1.20)
Equal	205 (49.5)	209 (50.5)	0.79 (0.67, 0.93)**	0.83 (0.67, 1.05)
Parity
1	511 (58)	370 (42)	1.32 (1.11, 1.57)**	0.59 (0.20, 1.74)
2	332 (53.4)	290 (46.6)	1.15 (0.95, 1.38)▪	0.67 (0.28, 1.61)
3	162 (57)	122 (43)	1.17 (0.95, 1.44)▪	0.73 (0.31, 1.74)
≥ 4	192 (50.8)	185 (49.2)	1	1
Number of previous ANC visits
<4	479 (59.4)	328 (40.6)	1.24 (1.07, 1.42)**	1.06 (0.89, 1.27)
≥ 4	574 (51.1)	549 (48.9)	1	1
Recent number of children
0–1	520 (58.2)	374 (41.8)	1	1
2–3	482 (54.2)	407 (45.8)	0.85 (0.75, 0.96)*	0.97 (0.51, 1.84)
≥ 4	187 (50.5)	183 (49.5)	0.76 (0.63, 0.89)**	0.70 (0.24, 2.06)
Past history of Stillbirth
Yes	22 (75.9)	7 (24.1)	1.76 (1.15, 2.68)**	1.18 (0.67, 2.03)
No	1174 (54.9)	964 (45.1)	1	1
Survival status of the recent child
Alive	1172 (54.7)	970 (45.3)	1	1
Died	25 (92.6)	2 (7.4)	6.28 (4.19, 9.41)***	2.90 (1.41, 5.97)**
Counselled during previous ANC visits
Yes	790 (50.6)	770 (49.4)	1	1
No	401 (67.6)	192 (32.4)	1.62 (1.42, 1.84)***	1.08 (0.85, 1.37)
Counselled during PNC visits
Yes	768 (50.4)	756 (49.6)	1	1
No	428 (66.9)	212 (33.1)	1.59 (1.40, 1.81)***	1.02 (0.81, 1.29)
Exclusive breastfeeding
Yes	1027 (53.5)	892 (46.5)	1	1
No	161 (68.2)	75 (31.8)	1.67 (1.40, 1.96)***	0.90 (0.63, 1.29)
Duration of breastfeeding
<24 months	1012 (69)	455 (31)	4.76 (4.01, 5.66)***	4.92 (3.95, 6.12)***
≥ 24 months	154 (23.6)	499 (76.4)	1	1
Decision maker for contraception
Husband	305 (66.9)	151 (33.1)	1	1
Wife	92 (53.5)	80 (46.5)	0.78 (0.61, 1.00)▪	1.31 (0.96, 1.79)
Jointly	795 (51.9)	738 (48.1)	0.67 (0.58, 0.77)***	1.07 (0.88, 1.29)
Discussion with husband
Yes	740 (48.5)	786 (51.5)	1	1
No	442 (71.4)	177 (28.6)	2.05 (1.81, 2.32)***	1.33 (1.10, 1.60)**
Husband encourages spacing
Yes	767 (50.4)	755 (49.6)	1	1
No	412 (68.1)	193 (31.9)	1.72 (1.51, 1.95)***	1.25 (1.05, 1.48)*
Modern contraceptive use
Yes	420 (39.7)	639 (60.3)	1	1
No	778 (70)	333 (30)	2.76 (2.44, 3.12)***	2.27 (1.94, 2.66)***
Plan to wait until current pregnancy
<24 months	60 (82.2)	13 (17.8)	2.56 (1.97, 3.34)***	1.72 (1.26, 2.35)**
≥ 24 months	1035 (54.2)	876 (45.8)	1	1
Desired number of children
1–5	348 (57)	263 (43)	1	1
≥ 6	656 (54.8)	542 (45.2)	0.95 (0.83, 1.09)	0.87 (0.74, 1.03)
Undecided	195 (53.9)	167 (46.1)	0.88 (0.73, 1.06)▪	0.84 (0.66, 1.07)
Mode of delivery for recent child
Spontaneous vaginal delivery	1116 (55.8)	884 (44.2)	1	1
Cesarean-section	51 (46.8)	58 (53.2)	0.79 (0.60, 1.05)▪	0.77 (0.55, 1.07)
Instrumental (forceps and vacuum)	31 (50.8)	30 (49.2)	0.95 (0.66, 1.38)	1.29 (0.82, 2.03)
Wealth status
Lowest	262 (60.9)	168 (39.1)	1	1
Second	234 (53.8)	201 (46.2)	0.90 (0.75, 1.07)	0.83 (0.66, 1.03)
Middle	249 (58.5)	177 (41.5)	1.01 (0.84, 1.21)	0.95 (0.76, 1.19)
Fourth	228 (52.7)	205 (47.3)	0.89 (0.74, 1.07)	0.88 (0.71, 1.11)
Highest	216 (50.1)	215 (49.9)	0.86 (0.71, 1.05)▪	0.94 (0.74, 1.19)
Exclusive breastfeeding* Contraception
No contraception*no exclusive breastfeeding				1.82 (1.17, 2.82)**
Plan to wait until current pregnancy*survival status of the recent child
<24 months*died				5.30 (1.06, 26.54)*
Theta				0.077 (0.011, 0.142)***
MHR				1.30 (1.11, 1.43)***
LR test of theta = 0
Chibar2(01)				41.64
Prob-hibar2				<0.001

Keys: *** = P<0.001,

** = P<0.01,

* = P<0.05, = P<0.20. AHR: Adjusted Hazard Ratio. CHR: Crude Hazard Ratio. CI: Confidence Interval. MHR: Median Hazard Ratio. LR: Likelihood Ratio. 1 = reference category.