Pregnancy and thrombosis risk for women without a history of thrombotic events: a retrospective study of the real risks

The statistical analysis was performed by the SAS for Windows 9.4 software platform [19] (SAS Institute Inc., NC, U.S.A.). The demographic and clinical data of the patients at the baseline were presented via descriptive statistics. Specifically, as the mean value and standard deviation (SD) and for reasons of completeness the median value and the values for 25% and 75% percentiles were also reported, for the qualitative data, frequencies and percentages were presented. Comparisons between two or more groups for the categorical parameters were performed using the chi-square test [20]. For the quantitative parameters (for example the women’s age) the Mann–Whitney U test was applied for comparison between two groups or the Kruskal–Wallis test when more than two groups were involved [19] respectively; as normality was not always certain. For paired analysis, i.e. to compare the PAT risk pre and post-partum, the Wilkoxon Signed Rank test was applied. The significance level (p-value) of the study was set to 0.05 and all tests were two-sided.

In terms of the baseline and treatment characteristics between women with pregnancy complications [Group-A: 445] and those with a prothrombotic tendency (thrombophilia, family history of VTE, other) [Group-C: 165] there were no significant differences in general; those undergoing IVF [Group-B: 132] were of a higher age (37.2 ± 5.1), one out of five had multiple pregnancy (30, 22.7%) and concomitantly received aspirin more frequently (30.3%). In terms of outcomes, again there were differences between the groups; specifically, in Group B the frequency of GVCs was almost twice as high in comparison with that of Group A and Group C, mainly driven by differences in IUGR and early pregnancy loss events. The characteristics of the study population are presented in Table 1.

The mean score antepartum was higher for women in Group B (3.3 in comparison with 1.9 and 2.0 in Group A and Group C respectively). Moreover, the risk score increased significantly postpartum for all Groups. The descriptive characteristics of the VTE risk score for each group, are presented in Table 2.

Afterwards, the differences in the risk scores between the three groups were compared by two methods: a) as an arithmetic score and b) as a qualitative measure (i.e. after being characterized as: low, intermediate and high).

A graphical representation of the results of the first approach (arithmetic score) is presented in Fig. 1. The statistical test indicated a significant difference in the scores between the three groups both before (p < 0.0001) and after birth (p < 0.0001).

The results of the investigation by score characterization that we conducted, are presented in Table 3. The data for risk antepartum and risk postpartum are presented separately. It is interesting to note that group B had a higher percentage of women at high risk (25.76% and 60.61% before and after delivery) compared to group A (18.88% and 28.31%) and group C (15.15% and 28.48%). A significant difference (p < 0.0001) was found in all comparisons between group B vs. group A or C.

Actually, group A and C have a lower distribution of the women in the high or intermediate risk group, while in group B 55.3% of the women were at intermediate or high risk before delivery and this percentage increased further after delivery (85.6%). The chi-square test also proved that there is a higher percentage of women at high or intermediate risk in group B compared to C before birth (55.3% vs.26.1% respectively, p < 0.0001, OR: 3.5, 95% CI: 2.2 – 5.7) and similarly after birth (85.6% vs. 56.4%, OR: 4.6, 95%CI: 2.6–8.2, p < 0.0001).

When comparing group B with the combined figures for groups A and C, we found that before birth the odds for Intermediate or high risk in group B were 3.8 times higher (95% CI: 2.6 – 5.6, p < 0.0001) than the odds in groups A and C combined (see Table 4), while after birth the OR was almost twice as high (7.0, 95% CI: 4.2–11.7, p < 0.0001).

In actual fact, the separation of women into the three groups studied is not the only one possible since women may have more than one reasons for being included in the study, and thus they could have potentially been included in more than one group. Within the source study [17], women were assigned to a group on the basis of one basic reason only. However multiple reasons could exist and the distribution of the women according to these reasons is depicted in Table 5. Actually 81.4% of the women in the study had a single reason while the remaining 18.6% had multiple reasons and thus these women had an increased thrombotic load. A graphical diagram depicting the overlapping of inclusion reasons in this study is depicted in Fig. 2.

With regard to the two major risk components that go into the calculation of the PATrisks score, that is a) pre-existing risk factors and b) obstetric risk factors, the distribution of the population studied according to the individual components is depicted in Table 6. A small percentage 11% had no risk factor, while 38% and 8% had only obstetric and pre-existing risk factors respectively; moreover, almost 43% of the women in the study, had simultaneously more than a single factor, that is, they had both pre-existing and obstetrics risk factors.

In total 12 (1.6%) out of 742 women included in our analysis experienced thrombotic events, whether pre- or post-partum, during the study period (see Table 1 [one woman had two thrombotic events]). The median value of the risk score for the women experiencing thrombosis was 2.5 (Q1-Q3: 0.5–2.5) before birth and this score rose to a median value of 3.5 (Q1-Q3: 2–3.5) after delivery. There was no significant difference in the score between the women who did and did not experience such events whether before (p = 0.4319) or after delivery (p = 0.4004) (see Fig. 3 for the relevant diagram), probably due to the anticoagulation prophylaxis given to all participating women. It is interesting to note that some women with high risk (score > 10) did not experience thrombotic events. Among women who did experience thrombosis, five had a low score, one had an intermediate score (i.e. 3) and the remaining 6 had a score higher than 4 (high risk).

The analysis of the six women with a low or intermediate score showed that two were in group A and four in group B while none were in group C. Out of these six women, one had a family history for thrombosis, one had a multiple pregnancy, one received blood transfusion and four had low risk thrombophilia; in addition, the one with the family history of VTE had a BMI of 42.1. Five out of the six thrombotic events were postpartum when no prophylaxis was applied, precisely because they had low risk score for VTE.

Of special interest is the distribution of high-risk thrombophilia observed in the study.

The figures indicating the presence of thrombophilia risk factors per Group and in Total are given in Table 7. Women with APLA who were in group B had a higher PATrisks score.

Out of the total study population of 742 women, 149 of them (20.1%) had high-risk thrombophilia (note that many women had multiple thrombophilias); specifically, there were 96 of them in group A, 26 in group B and 27 in group C, representing 21.6%, 19.7% and 16.4% out of each individual group population respectively. No statistically significant difference was found in the percentages of population with high-risk thrombophilia in the groups studied (p > 0.05 in all comparisons). Low-risk thrombophilia was observed in 83.7% of the total population, that is 82.3% in group A, 81.1% in group B and 89.7% in Group C; no significant difference was found between groups A and B (p = 0.7554); however, there was noticeable differences between groups A and C (OR: 1.9, 95% CI: 1.1—3.3, p = 0.0248) as well as between groups B and C (OR: 2.0, 95% CI: 1.0—4.0, p = 0.0338). However, the low risk factors with well-established association with VTE such us the heterozygous mutations for FV Leiden and FII 20,210 were found in 39.1% and 21.1% of the total population respectively (difference 18%, 95%CI: 13.3–22.6%, p < 0.0001). Finally, none of the thrombophilia risk factors presented in Table 7 was observed in 102 of the women (13.7% of the total population); these women were distributed as follows: 66 in group A (14.8% of the group population), 23 in group B (17.4%) and 13 in group C (7.9%).

In this post-hoc analysis, we retrospectively evaluated thrombotic risk in women who received LMWH based on clinical experience hoping to optimize their pregnancy outcome, so that we could investigate whether that practice was justified according to the PATrisks scoring system. The arithmetic mean score antepartum was higher for women in Group B (3.3) in comparison with that in Group A and Group C (1.9 and 2.0 respectively). This risk score increased significantly postpartum for all Groups. On the basis of the score characterization as high, intermediate and low, group B had a statistically significant higher percentage of women at high risk (25.76% and 60.61%, before and after delivery) compared to group A (18.88% and 28.31%) and group C (15.15% and 28.48%) in all comparisons. In relation to the major risk components that go into the calculation of the PATrisks score, a small percentage of women (11%) had no risk factor, while 38% and 8% had only obstetric and pre-existing risk factors respectively; moreover, almost 43% of women had more than a single risk factor type simultaneously. In total, 12 (1.6%) out of the 742 women included in our analysis experienced thrombotic events during the study. The median value of the risk score for the women experiencing thrombosis was 2.5 (Q1-Q3: 0.5–2.5) before birth and this score increased to a median value of 3.5 (Q1-Q3: 2–3.5) after delivery. No significant difference was found in the score between the women who did and did not experience such events either before (p = 0.4319) or after delivery (p = 0.4004).

Out of the total study population, 149 (20.1%) women had high risk thrombophilia; specifically 21.6% in group A, 19.7% in group B and 16.4% in group C. Absence of thrombophilia risk factors was observed in 102 women (13.7% of the total population), specifically in 14.8%, 17.4% and 7.9% of the women in groups A, B and C respectively.

Although an established VTE in pregnancy may be successfully treated with therapeutic doses of heparin, prevention is preferable to cure because of the high mortality and long-term morbidity associated with established disease [21, 22]. LMWHs can be used safely in pregnancy and in breastfeeding with no adverse effects on the fetus or the neonate. Medical prophylaxis against thrombosis starts during pregnancy and is generally continued for about six weeks following delivery due to the risk of thrombosis which peaks during the postpartum period. Because of their efficacy and good safety profile, the use of low molecular weight heparins for thromboprophylaxis in pregnancy has increased greatly in recent years. Their use has also increased in women at high risk of GVCs and for the optimization of ART outcomes in anticipation of the scientific evidence in favor of their use becoming available. Clinicians have proposed these treatments on the grounds of biological plausibility and on the basis of extrapolation from antiphospholipid syndrome [23]. There is also increasing evidence for the use of heparin in women with pregnancy complications mediated by the placenta (as can be seen from successful previous pregnancy outcomes), while this is not the case for women with thrombophilic defect alone [24].

In our cohort, administration of LMWH for VTE prevention was found to be both efficacious and safe; only 1.6% out of the total women population included in our analysis experienced thrombotic events (whether pre- or post-partum) during the study, and the prevalence of bleeding events was also low.

From our analysis we found that despite the fact that the administration of LMWHs in all groups was based on clinical experience, in one out of four women for Groups A & C and in more than half of women in Group B this was justified antepartum by the scoring tool since they were found to have intermediate or high risk for VTE. Additionally, the administration of LMWHs postpartum was justified to a higher extend; specifically, almost half of the women in Groups A & C and more than eight out of ten in Group B were found to have intermediate or high risk, after birth. It is interesting to note that, regardless of which group they had been allocated to, the number of women with intermediate or high score was almost twice as high postpartum (393) as antepartum (223).

With regard to the major risk components that go into the calculation of the PATrisks score only one out of ten women had no risk factors at all, while 38% and 8% of them had only obstetric or pre-existing risk factors respectively; moreover, almost 43% of the women, had both pre-existing and obstetrics risk factors simultaneously. Those findings also provide, at least partially, justification for the administration of LMWH in the population under study.

Robertson et al. [25] conducted a large systematic review to assess the overall relationship between all major thrombophilia and adverse pregnancy outcomes, which comprised 79 studies, primarily consisting of case–control studies. All inherited thrombophilia defects were associated with venous thromboembolic disease (VTE), homozygous/ heterozygous FV G1691A and FII G20210A were associated with pregnancy loss < 24 weeks [Odds ratio (OR) 2.71, 1.68, 2.49 respectively], heterozygous FV G1691A, FII G20210A, and PS with pregnancy loss > 24 weeks (OR 2.06, 2.66, 20.09), heterozygous FV G1691A and FII G20210A with PE (OR 2.19, 2.54), and heterozygous FV G1691A and FII G20210A with PA-VTE (OR: 4.70, 7.71).

That being so, in the PATrisks tool all the “high risk thrombophilias” (Table 7) are rated with 3 points, and from the “low risk thrombophilias” only the heterozygous mutations for FV Leiden and FII 20,210 are rated with 1 point. The “other risk thrombophilias” are not scored in the PATrisks tool because their association with the occurrence of VTE is not well established. Out of all the women in the study, one out of five had high risk thrombophilia and one out of ten had APLA; both conditions during pregnancy also provide a rationale for the use of thromboprophylaxis. Additionally, the low risk factors such us the heterozygous mutations for FV Leiden and FII 20,210 which were found in 39,1% and 21,1% of the total population respectively, again provides some rationale for primary prophylaxis from thrombotic events in the context of pregnancy and puerperium only if there are other preexisting or/and obstetrics risk factors.

In terms of pregnancy outcomes and despite the quite high prevalence of various thrombophilic disorders with known association with pregnancy complications, the frequency of GVCs in the pregnancies examined was low overall; the most frequently observed GVCs were preterm labor and IUGR.

In our days, assisted reproductive technologies (ART) are widely used in couples with fertility problems and low-molecular-weight heparin (LMWH) is routinely administered in various cases to improve ART outcomes. The live birth rate in Group B of our cohort was 97% which is quite high for women undergoing assisted reproduction techniques, indicating a possible favorable effect of LMWH administration in successful ART outcomes. Live birth rates were also high (higher than 99%) in both Groups A & C.

This post-hoc analysis also shares the limitations, as well as the advantages, of most observational studies. By its design, the study involves a broad range of everyday/routine clinical approaches, and there is limited specific focus for the inclusion in Groups of the participating women. Thus, biases of an unknown nature may have colored the results. This study does not feature the typical separation of patients into case and control groups—instead, the analysis involved all patients in an intention-to-treat manner. This list of limitations is by no means exhaustive. However, in the authors’ opinion, this study has the potential to capture the “in vivo” conditions of a common clinical obstetrics setting. We recognize that a control group of patients would strengthen the analysis. As we mentioned before, the study is retrospective, and it is a post hoc analysis of our previous work [17]. Pregnant women had received LMWH hoping to achieve a good pregnancy outcome. Their data were analyzed retrospectively to determine whether LMWH could provide to some of them an additional benefit as thromboprophylaxis, according to the PATRisks score which is based on RCOG guidelines. These guidelines have emerged from older, large studies which recognize the increased risk of VTE in pregnancy and postpartum [1‐3, 18]. In the future, we would like to study the prospective evaluation of the PATrisks score in pregnancy.