Pregnancy Loss and the Risk of Spontaneous Preterm Birth
Pregnancy Loss and the Risk of Spontaneous Preterm Birth
This large prospective cohort study of healthy nulliparous women with singleton pregnancies has demonstrated that women with previous pregnancy loss were at increased risk of spontaneous preterm birth, if they were managed by procedures involving cervical dilatation and curettage. Furthermore, a greater association between women with a history of two or three procedures involving cervical dilatation and curettage and spontaneous preterm birth, compared with women with only one previous cervical dilatation and curettage, was demonstrated (adjusted OR 2.32; 95% CI 1.88, 2.88 versus adjusted OR 1.66; 95% CI 1.14, 2.42). Overall, women with a single pregnancy loss did not have an increased risk of having any other of the adverse pregnancy outcomes examined. In contrast, two to four previous pregnancy losses were associated with an increased risk of having a pregnancy complicated by spontaneous preterm birth and/or placental abruption.
We then separated fetal loss into miscarriage and termination of pregnancy. Although one previous miscarriage overall was not associated with an increased risk of spontaneous preterm birth, when women were managed with procedures involving cervical dilatation and curettage, a significant risk of spontaneous preterm birth was observed. Women with one previous termination of pregnancy had an overall increased risk of spontaneous preterm birth but again this risk was restricted to when the previous termination of pregnancy was managed surgically. These associations were not gestation dependent, because the results remained unchanged when analyses were confined to <10 weeks' gestation.
Women with two previous miscarriages were at an increased risk of either spontaneous preterm birth or PPROM. This may reflect an increased incidence of undiagnosed medical problems associated with recurrent miscarriage such as anti-phospholipid syndrome (Rai et al., 1997) or reflect an underlying predisposition towards poor placentation (Bose et al., 2006).
The strengths of our study are that detailed information about pregnancy outcomes were collected prospectively with pregnancy outcome data available in 99% of participants. Pregnancy outcome was assigned according to pre-specified criteria and stringent data monitoring protocols ensured the quality of the data. Although every effort was made to record accurate previous pregnancy loss information and management by the trained SCOPE midwives using a detailed pregnancy loss proforma, it was not feasible to confirm the history and management of previous pregnancy loss by hospital records. This may have introduced recall bias. As pregnancy loss is such an important event for mothers, it is unlikely that this information would be prone to recall bias (Hewson and Bennett, 1987; Githens et al., 1993; Yawn et al., 1998). Furthermore, previous pregnancy loss data were recorded at 15 weeks' gestation prior to the occurrence of any of the observed adverse pregnancy outcomes. The mechanisms to explain the association between cervical dilatation/curettage and spontaneous preterm birth are unknown. Postulated mechanisms include damage to cervical tissues by artificial dilatation of the cervix or alteration in the expression of genes involved in collagenolysis and inflammation following disruption of the endometrium/myometrium during curettage (Sooranna et al., 2005). As women with three or more miscarriages or three or more terminations of pregnancy were excluded from the SCOPE study due to an accepted assumption of increased risk and therefore increased surveillance, this study cannot assess the potential 'dose' effect beyond two losses but the data suggest that an increased risk is already present after two pregnancy losses (termination of pregnancy or miscarriage) and comparable with the increased risk reported with higher degrees of loss (Thom et al., 1992). In this study, we conducted analyses of multiple exposures, including the analysis of interaction terms, to examine several outcomes in this cohort. Given the multiple comparisons, there is the possibility of rejecting null hypotheses incorrectly. Non-surgical methods of management of miscarriage and termination of pregnancy would have included medical methods as described in the methods section or occasionally expectant management. As management was classified simply as surgical or non-surgical, we were unable to explore whether any differences occurred between those women managed expectantly or medically.
Published data regarding the association between previous pregnancy loss and adverse pregnancy outcomes in subsequent pregnancies are limited, with conflicting results (van Oppenraaij et al., 2009; Virk et al., 2007). Some studies have reported that miscarriage is associated with an increased risk of preterm delivery and PPROM (Swingle et al., 2009; Buchmayer et al., 2004) and SGA (Basso et al., 1998; Bhattacharya et al., 2008), whereas others have not (Schoenbaum et al., 1980; de Haas et al., 1991; Ekwo et al., 1993; Hammoud et al., 2007). Similarly, conflicting evidence exists regarding previous termination of pregnancy and subsequent adverse pregnancy outcomes (Hogue et al., 1983; Pickering and Forbes, 1985; Atrash and Hogue, 1990; Lang et al., 1996; Zhou et al., 1999; Ancel et al., 2004; Moreau et al., 2005; Raatikainen et al., 2006). Few studies have examined whether the mode of management of miscarriage or termination of pregnancy is relevant (Lohmann-Bigelow et al., 2007; Virk et al., 2007).
Prospective, well-conducted studies are lacking and many of the retrospective studies have small numbers and are of poor quality with a significant bias making definitive conclusions and comparisons difficult. In our study, the association between one previous miscarriage and spontaneous preterm birth was confined to women managed surgically. Our study does not support an association between one previous miscarriage and other pregnancy complications as shown by others (Buchmayer et al., 2004; Bhattacharya et al., 2008) but concurs with a previously reported association between two previous miscarriages and an increased risk of spontaneous preterm delivery (Buchmayer et al., 2004) and PPROM but not SGA, as demonstrated in other studies (Basso et al., 1998; Buchmayer et al., 2004). Our results differ from those reported by Lohmann-Bigelow et al. who found no association between previous dilatation and curettage and preterm delivery, pre-eclampsia and placental abruption and miscarriage in their retrospective study. These findings may be explained by the heterogeneous population of multiparous women included in this study or the fact that pregnancy outcomes were obtained retrospectively, in contrast with the prospective nature of the SCOPE cohort of nulliparous women (Lohmann-Bigelow et al., 2007). Whilst this study is consistent with those that have found an association between women with one previous termination of pregnancy and an increased risk of preterm delivery, we did not confirm the previously reported association with PPROM (Ancel et al., 2004). The SCOPE database allowed us to access many potential confounding factors such as drug use. Women recruited to the SCOPE cohort were healthy nulliparous women with no known medical conditions thought to influence the risk of preterm birth, SGA or pre-eclampsia. Nonetheless, there is the possibility that undiagnosed medical conditions such as untested thyroid abnormalities may act as further potential confounding factors.
In conclusion, women with either one previous miscarriage or previous termination managed non-surgically were not significantly associated with an increased risk of spontaneous preterm birth. In contrast, when management of miscarriage or termination included cervical dilatation and curettage, a significant association was demonstrated for spontaneous preterm birth. A significant association between women with multiple (two to four) pregnancy losses and spontaneous preterm birth and/or placental abruption was also demonstrated. Despite demonstrating a significant association between pregnancy loss managed by dilatation and curettage and spontaneous preterm birth, an interpretation of any causal effect of dilatation and curettage is not possible and further studies are needed to examine this. Further research is also required to confirm our findings in other populations and to determine whether non-surgical management of miscarriage or termination of pregnancy should be advocated over surgical treatment.
Discussion
This large prospective cohort study of healthy nulliparous women with singleton pregnancies has demonstrated that women with previous pregnancy loss were at increased risk of spontaneous preterm birth, if they were managed by procedures involving cervical dilatation and curettage. Furthermore, a greater association between women with a history of two or three procedures involving cervical dilatation and curettage and spontaneous preterm birth, compared with women with only one previous cervical dilatation and curettage, was demonstrated (adjusted OR 2.32; 95% CI 1.88, 2.88 versus adjusted OR 1.66; 95% CI 1.14, 2.42). Overall, women with a single pregnancy loss did not have an increased risk of having any other of the adverse pregnancy outcomes examined. In contrast, two to four previous pregnancy losses were associated with an increased risk of having a pregnancy complicated by spontaneous preterm birth and/or placental abruption.
We then separated fetal loss into miscarriage and termination of pregnancy. Although one previous miscarriage overall was not associated with an increased risk of spontaneous preterm birth, when women were managed with procedures involving cervical dilatation and curettage, a significant risk of spontaneous preterm birth was observed. Women with one previous termination of pregnancy had an overall increased risk of spontaneous preterm birth but again this risk was restricted to when the previous termination of pregnancy was managed surgically. These associations were not gestation dependent, because the results remained unchanged when analyses were confined to <10 weeks' gestation.
Women with two previous miscarriages were at an increased risk of either spontaneous preterm birth or PPROM. This may reflect an increased incidence of undiagnosed medical problems associated with recurrent miscarriage such as anti-phospholipid syndrome (Rai et al., 1997) or reflect an underlying predisposition towards poor placentation (Bose et al., 2006).
The strengths of our study are that detailed information about pregnancy outcomes were collected prospectively with pregnancy outcome data available in 99% of participants. Pregnancy outcome was assigned according to pre-specified criteria and stringent data monitoring protocols ensured the quality of the data. Although every effort was made to record accurate previous pregnancy loss information and management by the trained SCOPE midwives using a detailed pregnancy loss proforma, it was not feasible to confirm the history and management of previous pregnancy loss by hospital records. This may have introduced recall bias. As pregnancy loss is such an important event for mothers, it is unlikely that this information would be prone to recall bias (Hewson and Bennett, 1987; Githens et al., 1993; Yawn et al., 1998). Furthermore, previous pregnancy loss data were recorded at 15 weeks' gestation prior to the occurrence of any of the observed adverse pregnancy outcomes. The mechanisms to explain the association between cervical dilatation/curettage and spontaneous preterm birth are unknown. Postulated mechanisms include damage to cervical tissues by artificial dilatation of the cervix or alteration in the expression of genes involved in collagenolysis and inflammation following disruption of the endometrium/myometrium during curettage (Sooranna et al., 2005). As women with three or more miscarriages or three or more terminations of pregnancy were excluded from the SCOPE study due to an accepted assumption of increased risk and therefore increased surveillance, this study cannot assess the potential 'dose' effect beyond two losses but the data suggest that an increased risk is already present after two pregnancy losses (termination of pregnancy or miscarriage) and comparable with the increased risk reported with higher degrees of loss (Thom et al., 1992). In this study, we conducted analyses of multiple exposures, including the analysis of interaction terms, to examine several outcomes in this cohort. Given the multiple comparisons, there is the possibility of rejecting null hypotheses incorrectly. Non-surgical methods of management of miscarriage and termination of pregnancy would have included medical methods as described in the methods section or occasionally expectant management. As management was classified simply as surgical or non-surgical, we were unable to explore whether any differences occurred between those women managed expectantly or medically.
Published data regarding the association between previous pregnancy loss and adverse pregnancy outcomes in subsequent pregnancies are limited, with conflicting results (van Oppenraaij et al., 2009; Virk et al., 2007). Some studies have reported that miscarriage is associated with an increased risk of preterm delivery and PPROM (Swingle et al., 2009; Buchmayer et al., 2004) and SGA (Basso et al., 1998; Bhattacharya et al., 2008), whereas others have not (Schoenbaum et al., 1980; de Haas et al., 1991; Ekwo et al., 1993; Hammoud et al., 2007). Similarly, conflicting evidence exists regarding previous termination of pregnancy and subsequent adverse pregnancy outcomes (Hogue et al., 1983; Pickering and Forbes, 1985; Atrash and Hogue, 1990; Lang et al., 1996; Zhou et al., 1999; Ancel et al., 2004; Moreau et al., 2005; Raatikainen et al., 2006). Few studies have examined whether the mode of management of miscarriage or termination of pregnancy is relevant (Lohmann-Bigelow et al., 2007; Virk et al., 2007).
Prospective, well-conducted studies are lacking and many of the retrospective studies have small numbers and are of poor quality with a significant bias making definitive conclusions and comparisons difficult. In our study, the association between one previous miscarriage and spontaneous preterm birth was confined to women managed surgically. Our study does not support an association between one previous miscarriage and other pregnancy complications as shown by others (Buchmayer et al., 2004; Bhattacharya et al., 2008) but concurs with a previously reported association between two previous miscarriages and an increased risk of spontaneous preterm delivery (Buchmayer et al., 2004) and PPROM but not SGA, as demonstrated in other studies (Basso et al., 1998; Buchmayer et al., 2004). Our results differ from those reported by Lohmann-Bigelow et al. who found no association between previous dilatation and curettage and preterm delivery, pre-eclampsia and placental abruption and miscarriage in their retrospective study. These findings may be explained by the heterogeneous population of multiparous women included in this study or the fact that pregnancy outcomes were obtained retrospectively, in contrast with the prospective nature of the SCOPE cohort of nulliparous women (Lohmann-Bigelow et al., 2007). Whilst this study is consistent with those that have found an association between women with one previous termination of pregnancy and an increased risk of preterm delivery, we did not confirm the previously reported association with PPROM (Ancel et al., 2004). The SCOPE database allowed us to access many potential confounding factors such as drug use. Women recruited to the SCOPE cohort were healthy nulliparous women with no known medical conditions thought to influence the risk of preterm birth, SGA or pre-eclampsia. Nonetheless, there is the possibility that undiagnosed medical conditions such as untested thyroid abnormalities may act as further potential confounding factors.
In conclusion, women with either one previous miscarriage or previous termination managed non-surgically were not significantly associated with an increased risk of spontaneous preterm birth. In contrast, when management of miscarriage or termination included cervical dilatation and curettage, a significant association was demonstrated for spontaneous preterm birth. A significant association between women with multiple (two to four) pregnancy losses and spontaneous preterm birth and/or placental abruption was also demonstrated. Despite demonstrating a significant association between pregnancy loss managed by dilatation and curettage and spontaneous preterm birth, an interpretation of any causal effect of dilatation and curettage is not possible and further studies are needed to examine this. Further research is also required to confirm our findings in other populations and to determine whether non-surgical management of miscarriage or termination of pregnancy should be advocated over surgical treatment.