Advertisement

Are infertile patients at greater risk of cesarean delivery?

      Over the last 30 years cesarean section delivery (CSD) rates have been steadily rising in the U.S., and around the globe. Current rates approximate 32% in the general population, and about 45% of infertile women who are treated with assisted reproductive technology (ART) are delivered by CSD. Why? Do clinicians have a lower threshold for delivering so-called precious in vitro fertilization (IVF) pregnancies by CSD electively? Are infertile patients inherently at higher risk for cesarean delivery than fertile couples? Could infertility treatments be associated with higher rates of CSD? Are there other factors to consider?
      In this issue, Stern et al. (
      • Stern J.E.
      • Liu C.
      • Cabral H.J.
      • Richards E.G.
      • Coddington C.C.
      • Missmer S.A.
      • et al.
      Factors associated with the increased odds of cesarean delivery in ART pregnancies.
      ) wrestle with the complexity of examining factors associated with increased risk of CSD. A plethora of investigators have provided evidence that women who receive ART treatment are at increased risk of perinatal complications and their sequelae. For example, a recent meta-analysis of 50 studies examined information from 161,370 ART-conceived singleton pregnancies and 2,280,241 pregnancies of fertile women who conceived spontaneously (
      • Qin J.
      • Liu X.
      • Sheng X.
      • Wang H.
      • Gao S.
      Assisted reproductive technology and the risk of pregnancy-related complications and adverse pregnancy outcomes in singleton pregnancies: a meta-analysis of cohort studies.
      ). It found that ART exposure is associated with increased odds of CSD (odds ratio 1.58; 95% confidence interval [CI] 1.48–1.70), but the results were heterogeneous. In addition, most of the reviewed studies had compared women who received ART treatment to women who were fertile and conceived spontaneously; (i.e., few considered the influence of sub-fertility and its correlates).
      We were able to find two relatively small studies that compared infertility group to infertility group by using third party reproduction (women who conceived with either donor eggs or gestational carriers) which allowed for a more direct assessment and eliminated some variables. The first is a Swedish study by Elenis et al. (
      • Elenis E.
      • Svanberg A.S.
      • Lampic C.
      • Skalkidou A.
      • Akerud H.
      • Sydsjo G.
      Adverse obstetric outcomes in pregnancies resulting from oocyte donation: a retrospective cohort case study in Sweden.
      ) who compared (group A) 76 egg donors versus (group B) 63 women who were infertile and received IVF treatment using autologous eggs versus (group C) 150 fertile women who conceived spontaneously. They found increased odds of hypertensive disorders, oligohydramnios, postpartum hemorrhage, retained placenta as well as non-emergency, and emergency CSD among infertile women compared to those who either conceived spontaneously or received IVF treatment using their own eggs. The donor oocyte recipients group, had increased odds of non-emergency CSD (adjusted odds ratio [aOR] 5.13; 95% CI 2.00–13.17) compared to fertile controls, although there was no difference when comparing donor oocyte recipients to women who received IVF treatment with autologous oocytes (group C) (aOR 1.82; 95% CI 0.71–4.66). By contrast, the odds of emergency CSD were significantly increased among donor recipients (aOR 15.98; 95% CI 3.27–78.23). In the second study, Woo et al. (
      • Woo I.
      • Hindoyan R.
      • Landay M.
      • Ho J.
      • Ingles S.A.
      • McGinnis L.K.
      • et al.
      Perinatal outcomes after natural conception versus in vitro fertilization (IVF) in gestational surrogates: a model to evaluate IVF treatment versus maternal effects.
      ) investigated the perinatal outcomes of 124 couples who used gestational carriers versus the outcomes of their prior pregnancies that were conceived spontaneously (i.e., historical controls). There was no difference in emergency CSD rates, however, the elective CSD rate was nearly threefold higher in women who used commissioned embryos than among women who conceived spontaneously (16% vs. 6%). In both of these cases, there is a higher elective cesarean section rate suggesting a role for precious pregnancies or maternal anxiety prompting elective CSD in infertile populations.
      The report by Stern and colleagues (
      • Stern J.E.
      • Liu C.
      • Cabral H.J.
      • Richards E.G.
      • Coddington C.C.
      • Missmer S.A.
      • et al.
      Factors associated with the increased odds of cesarean delivery in ART pregnancies.
      ) is based on a retrospective analysis of information from 173,130 primiparous women who delivered in Massachusetts from 2004–2010. They examined data from approximately 90% of the ART-conceived pregnancies in the state of Massachusetts by using Massachusetts Outcome Study of Assisted Reproductive Technology, a database created by a five-phase linkage algorithm that combined the Society for Assisted Reproductive Technology Clinic Outcome Reporting System and Massachusetts birth and hospital discharge records in the Massachusetts Outcome Study of Assisted Reproductive Technology. The cohort was divided into four groups: 5,768 ART-treated women; 1,627 subfertile women with non-ART medically assisted reproduction; 1,030 subfertile women who did not receive ART; and 164,705 fertile women. The results, not surprisingly, are similar to other regional studies that showed higher odds of CSD when comparing women who were exposed to ART to fertile women who conceived spontaneously. Yet, there was no difference in the odds of CSD among subfertile women who either received medically assisted reproduction or were untreated. Subfertility, per se, or its correlates, is therefore associated with increased odds of CSD in this sample.
      The authors of each of the studies just described acknowledged concern about residual bias (i.e., that which remains even after adjusting for potentially confounding factors). For example, the gestational carriers included in the study by Woo et al. (
      • Woo I.
      • Hindoyan R.
      • Landay M.
      • Ho J.
      • Ingles S.A.
      • McGinnis L.K.
      • et al.
      Perinatal outcomes after natural conception versus in vitro fertilization (IVF) in gestational surrogates: a model to evaluate IVF treatment versus maternal effects.
      ) were at increased risk of gestational diabetes, hypertension, and placenta previa, making it difficult to determine whether these factors or the fact that they were gestational carriers was responsible for their increased risk of CSD. To account for the influence of potentially confounding factors, Stern and colleagues (
      • Stern J.E.
      • Liu C.
      • Cabral H.J.
      • Richards E.G.
      • Coddington C.C.
      • Missmer S.A.
      • et al.
      Factors associated with the increased odds of cesarean delivery in ART pregnancies.
      ) developed logistic regression models first by including maternal demographic parameters and medical history, and second by additionally adjusting for pregnancy complications and delivery characteristics. To reduce the number of variables considered, the authors used a backwards model selection strategy. It is difficult to differentiate direct causal relationships with these commonly used, but limited tools. Especially for common outcomes with a plethora of risk factors like CSD. Namely, because logistic regression assumes that all important risk factors are included in the model, which is unlikely when we do not know all of the factors that are responsible for CSD. It is also possible that the relationship between ART and cesarean delivery depends on the value of another independent variable (e.g., age), yet interaction terms do not appear to have been considered. Odds ratios are also optimistically biased compared to the risk or hazard ratio. Misclassification by way of missing data or measurement error in administrative records are also capable of explaining the estimated 15% to 29% higher odds of CSD linked to ART treatment when compared to fertile women who conceive spontaneously. All of these factors may be influential.
      The key question about the report by Stern et al. (
      • Stern J.E.
      • Liu C.
      • Cabral H.J.
      • Richards E.G.
      • Coddington C.C.
      • Missmer S.A.
      • et al.
      Factors associated with the increased odds of cesarean delivery in ART pregnancies.
      ) is whether key risk factors are left out of the logistic regression model, because if they are associated with ART or its correlates (i.e., maternal age and medical history), then adjustment for these factors can distort or fully explain an estimated association between ART and CSD. The major strength of the authors' approach is their capacity to classify couples as medically treated ART, subfertility treated, untreated infertile or fertile. That is, the study design enabled examination of evidence consistent with fertility problems per se increasing the odds of cesarean section delivery apart from the ART-treatment that was administered to combat such problems. Because untreated subfertility was associated with increased odds of CSD when compared to fertile patients, and because ART exposure is not associated with cesarean section delivery when examined only among couples who had sub- or in-fertility problems, ART does not appear to be strongly associated with increased risk of CSD. Instead, fertility problems, or other correlates not included in the logistic model, may drive the association. Indeed, confounding by indication is the most stubborn type of bias, because no statistical adjustment strategy can fully resolve it (
      • Bosco J.L.
      • Silliman R.A.
      • Thwin S.S.
      • Geiger A.M.
      • Buist D.S.
      • Prout M.N.
      • et al.
      A most stubborn bias: no adjustment method fully resolves confounding by indication in observational studies.
      ).
      In conclusion, infertility (with or without treatment) is associated with increased odds of CSD compared to women who are fertile who conceive spontaneously, and this might explain the relationship between ART and CSD observed by Stern and colleagues (
      • Stern J.E.
      • Liu C.
      • Cabral H.J.
      • Richards E.G.
      • Coddington C.C.
      • Missmer S.A.
      • et al.
      Factors associated with the increased odds of cesarean delivery in ART pregnancies.
      ). Even so, their ability to link the Society for Assisted Reproductive Technology Clinic Outcome Reporting System database to state vital records, despite some limitations, is an important demonstration of leveraging available resources to learn more about complex pregnancy disorders. The results of this study will indeed aid those of us who are clinicians in counseling our patients regarding their risks of pregnancy and delivery complications, and about our limited capacity to estimate these associations.

      References

        • Stern J.E.
        • Liu C.
        • Cabral H.J.
        • Richards E.G.
        • Coddington C.C.
        • Missmer S.A.
        • et al.
        Factors associated with the increased odds of cesarean delivery in ART pregnancies.
        Fertil Steril. 2018; 110: 106-113
        • Qin J.
        • Liu X.
        • Sheng X.
        • Wang H.
        • Gao S.
        Assisted reproductive technology and the risk of pregnancy-related complications and adverse pregnancy outcomes in singleton pregnancies: a meta-analysis of cohort studies.
        Fertil Steril. 2016; 105: 73-85.e6
        • Elenis E.
        • Svanberg A.S.
        • Lampic C.
        • Skalkidou A.
        • Akerud H.
        • Sydsjo G.
        Adverse obstetric outcomes in pregnancies resulting from oocyte donation: a retrospective cohort case study in Sweden.
        BMC Pregnancy Childbirth. 2015; 15: 247
        • Woo I.
        • Hindoyan R.
        • Landay M.
        • Ho J.
        • Ingles S.A.
        • McGinnis L.K.
        • et al.
        Perinatal outcomes after natural conception versus in vitro fertilization (IVF) in gestational surrogates: a model to evaluate IVF treatment versus maternal effects.
        Fertil Steril. 2017; 108: 993-998
        • Bosco J.L.
        • Silliman R.A.
        • Thwin S.S.
        • Geiger A.M.
        • Buist D.S.
        • Prout M.N.
        • et al.
        A most stubborn bias: no adjustment method fully resolves confounding by indication in observational studies.
        J Clin Epidemiol. 2010; 63: 64-74

      Linked Article