This study demonstrates that the highest risk for a spontaneous abortion was following another spontaneous abortion, or after a short IOI <=14 months irrespective of the preceding outcome. After a stillbirth, the IOIs <=14 months were associated with a higher risk of another stillbirth or a spontaneous abortion. Having a prior neonatal death was associated with an increased risk of another neonatal death or a non-live outcome for intervals up to 26 months relative to 27-50 months. The risk of a stillbirth after a neonatal death was also significantly higher for IOI’s >72 months, suggesting a U-shaped relationship where the risk decreased up to 27 months, but then increased again after 72 months. Overall, following an adverse outcome, a short IOI was associated with a risk of another adverse outcome in the subsequent pregnancy. Further research that appropriately measures gestational age and IPI is needed to help guide family planning interventions.
A recent study from rural Bangladesh has demonstrated that incorporating post-partum family planning into the maternal and newborn care packages can significantly improve birth spacing and reduce the risk of preterm birth [adjusted RR = 0.79; 95% CI = 0.63-0.99)] . Given the relative lack of data from low income countries, it is understandable that there are few protocols for family planning counseling for women who have experienced non-live birth outcomes. The expert recommendation to WHO has been to wait six months after a spontaneous abortion before conceiving again . The experts noted, however, that this recommendation was based on only one large study (n=258,108 women) in Latin America from two countries (Argentina and Uruguay) that used hospital records. This study was also unable to distinguish between spontaneous and induced abortions. Since then, a 2016 study in India by Chadna et al. has supported this recommendation . This India study found that “women who conceive between 6-12 months of an initial miscarriage have better outcomes and lower complication rates in their subsequent pregnancy.”
Most of the available literature on the recommendation for how long to wait after an adverse event before conceiving again come from high income countries and focus on waiting time after a miscarriage. Some results from high income countries suggest that conception within a short interval (6 months) after an abortion does not result in adverse pregnancy outcomes [34,35,36,37,38]. A recent review of 16 studies also led to the same conclusion .
In contrast, results from low- and middle-income countries [Latin America, Bangladesh (Matlab), India] and our current analysis of Bangladesh (Sylhet) indicates that short intervals after a stillbirth, neonatal death or spontaneous abortion, are associated with increased risk of adverse pregnancy outcomes. A 2010 study in Matlab, Bangladesh, also showed that the maternal mortality risk was “higher for pregnancies that ended in induced abortion, miscarriage, or stillbirth, compared to those that resulted in a live birth” (odds ratios 4.2, 2.0, and 17.4, respectively) . The study concluded that improved management of these outcomes was needed in order to reduce the maternal mortality associated with them. A Bangladesh study , based on 125,720 births, found that “very short intervals (< 15 months) following stillbirths and miscarriages were associated with significantly increased risks of early neonatal mortality (RR =1.87, p<0.001, and RR=1.48, p<0.01, respectively)” compared with inter-outcome intervals of 36-59 months that followed pregnancies in which the infant was born alive. The study found that other intervals after pregnancies that began with non-live birth outcomes also were associated with significantly higher mortality risk and the and the effects “did not vary much by interval length.” . The authors of this study recommended that women experiencing these types of outcomes be advised to wait at least six months before becoming pregnant again.
Another Bangladesh study, based on 10,453 pregnancies found that, compared with inter-pregnancy intervals (IPIs) of 6-12 months, pregnancies that were conceived <3 months after a miscarriage were more likely to result in a live birth . However, the study also found that very short IPIs of <3 months following a miscarriage were associated with significantly higher risk of late neonatal mortality for the infant born at the end of the IPI (adjusted hazard ratio (HR) 1.74, 1.06 to 2.84). The Conde-Agudelo study  found that in Latin America, compared with post-abortion (spontaneous and induced) IPIs of 18-23 months, inter-pregnancy intervals shorter than 6 months were significantly associated with increased risk of low birth weight, very low birth weight, preterm delivery, very preterm delivery, and adverse maternal outcomes in the next pregnancy.
Somewhat in contrast to these findings, a study from Bangladesh, based on 66,759 pregnancy outcomes, found that “if the preceding pregnancy ended in a miscarriage or stillbirth, there is an elevated risk that the index pregnancy will end with the same outcome, regardless of the amount of time since the previous pregnancy ended” . The differences in the results between studies conducted in higher and lower resource settings, as well as within low-resource settings, may be due to contextual differences such as the availability of good healthcare facilities which helps ensure that such high risk pregnancies are more closely monitored and cared for. For example, DaVanzo’s data were drawn from Matlab, Bangladesh, an area having higher quality care, compared to services in rural, northeast Bangladesh. Differences in nutritional status, risk of infections and genetic predispositions may also be important factors.
Most of the studies in the literature have investigated the effect of intervals defined as inter-pregnancy or inter-birth depending on the available data [2, 5, 9, 10, 14, 28, 42, 43]. Considering that we did not have data on gestational age, our intervals were defined according to DaVanzo et al.  who studied pregnancy spacing and infant and child mortality. Despite the difference in the target age groups, we have found similar results to DaVanzo et al. They undertook analysis with and without (i.e. using IOIs) having data on gestational age, and the results were similar. The shortest and longest intervals in our study were associated with higher risks of adverse perinatal outcomes.
Evidence of effect modification of the interval by maternal age has been found where the risk of preterm delivery for older mothers (aged >34 years) was lower at shorter IPIs (<11 months) but higher at longer IPI’s (11-23 months) in comparison to mothers aged 25-34 years . In our analysis, we found no evidence of interaction between IPI and mother’s age. Our results have also provided further evidence of clustering of adverse outcomes of the same type within the same mother. An in-depth discussion of clustering of neonatal deaths in this study population was given previously , and a similar discussion in an Indian setting was given by Williams et al. .
Mechanisms through which intervals affect outcomes
We identified two possible mechanisms that might explain how short intervals may affect pregnancy outcomes: folate depletion and vertical transmission of infection. We discuss these mechanisms because there is evidence in the literature regarding these mechanisms’ effects on perinatal outcomes (not just infant child outcomes) and there is also evidence on Bangladeshi women’s health conditions as they relate to these mechanisms.
Folate depletion has been hypothesized as a mechanism through which short IPIs are associated with adverse perinatal outcomes. If pregnancy occurs three to four months after a birth, before sufficient repletion of folate resources, the risk of another miscarriage may increase. A review of seventeen studies in low, middle and high income countries concluded that “strong evidence exists that folate depletion occurs in women during the first three to four months postpartum.”. Studies in rural Bangladesh have found folate deficiency in early pregnancy and among married, nulliparous women [46, 47]. One study of over 11,000 women examined folate levels, miscarriage, and stillbirth, and found that, after adjustment for confounders, compared to women without supplemental folate intake, those in the highest category of intake had a reduced relative risk of spontaneous abortion of 0.80 (p=0.001 95% CI 0.71, 0.90) and reduced risk of stillbirth 0.55 (0.30, 1.00) (p=0.06) .
With respect to pregnancy intervals, one study found that mean erythrocyte and serum folate levels were significantly lower among women with short IPIs (6 months or less) compared to women with longer intervals of 18-24 months (erythrocyte p=0.002; serum folate p =0.00001) . A second study did not find an association between IPI and folic acid deficiency. However, this study examined only two categories of IPIs (≤30 and > 30 months), and did not examine shorter intervals such as less than 6, 6-11,12-17, 18-23, or 24-29 months.
A review of causal mechanisms concluded that there is evidence to support the hypothesis that folate depletion constitutes a hypothetical mechanism that explains the increased risk of adverse perinatal outcomes in women with short pregnancy intervals (43). However, this review focused on birth weight and small for gestational age outcomes, rather than neonatal death, stillbirth, and miscarriage.
With respect to vertical transmission of infections, women infected with bacterial, fungal, or viral organisms may harbor the organisms at a site from which the newly-conceived fetus can be infected. One study noted that, in theory, the risk of infection of the fetus could be higher for women with short intervals because they have less time to recover between pregnancies . Studies have found high prevalence of reproductive tract and sexually transmitted infections in Bangladesh . One analysis concluded that cytomegalovirus, the most frequent congenital infection globally, is endemic in Bangladesh [52, 53]. It found that among 420 pregnant Bangladeshi women, the prevalence of cytomegalovirus IgG antibody was 66.7% by the age of 15-20 years, and 71.4% in the age group of 26-30 years. Studies have found an association between primary cytomegalovirus infection, miscarriage and fetal, neonatal and infant deaths [54, 55]. One study of 3,461 women assessed the effect of the interval between births on the risk of congenital CMV infection . It found that women who seroconverted between deliveries ≤ 24 months apart had a four-fold higher risk of delivering a congenitally infected baby than women who seroconverted between deliveries >24 months apart (OR, 4.3; 95% CI, 1.4-14.2). This study, however, did not adjust for confounding factors such as maternal age, race, and socioeconomic status.
Study strengths and limitations
The first strength of our study is that our results are based on a large sample size and the pregnancy history survey was conducted as part of a household survey rather than in a hospital setting. Thus, our study was not likely to suffer from selection bias resulting from sampling only those who seek care at formal institutions. Per criteria set forth for assessing the quality of observational studies on pregnancy spacing , we examined five categories of IOI; controlled for 10 potential confounders, and we had only 2.3% missing outcomes.
The main limitation of our study is that CHWs relied on pregnancy history recall rather than obtaining data from medical records. Thus there were no gestational ages which would have enabled us to investigate the inter-pregnancy interval effect. The errors in the recall of timing of events may explain why there are reported outcomes other than spontaneous abortion within the 6 months IOIs, which is not possible. There may also be some misclassification bias where early neonatal deaths were reported as stillbirths and the other way round, and similarly for spontaneous abortions and induced abortion. However, to reduce errors in recall of dates, the CHWs in our study carried a list of dates for historical events that happened in recent previous months or years and they asked mothers to use those dates as reference as well. Furthermore, a conversion calendar was used so that if women reported a date using a local calendar, a corresponding western calendar date could be recorded.