By Jeffrey T. Jensen, MD, MPH
Leon Speroff Professor and Vice Chair for Research, Department of Obstetrics and Gynecology, Oregon Health & Science University, Portland
Dr. Jensen reports he is a consultant for and receives grant/research support from HRA Pharma, Bayer Healthcare, Merck, Agile Pharm, Population Council, AbbVie, Evofem, and ContraMed.
SYNOPSIS: In a meta-analysis, a history of dilation and curettage for management of miscarriage or termination of pregnancy was associated with an increased risk of preterm birth in a subsequent pregnancy but the association is weak and most likely explained by confounding.
SOURCE: Lemmers M, et al. Dilatation and curettage increases the risk of subsequent preterm birth: A systematic review and meta-analysis. Hum Reprod 2016;31:34-45.
Since dilation and curettage (D&C) is one of the most frequently performed procedures in reproductive-aged women, any effect of the procedure on outcomes of subsequent pregnancies is important. To evaluate whether D&C increases the risk of subsequent preterm birth, the authors performed a systematic review and meta-analysis of previously published studies. Since no randomized studies exist, they searched the medical literature and selected cohort and case-control studies comparing subsequent preterm birth in women who previously had been managed surgically for first trimester miscarriage or termination of pregnancy to a control group of women without a history of D&C (predominantly suction curettage). From more than 2100 citations identified, only 21 studies met the inclusion criteria. These reported outcomes of 1,853,017 women; 77,231 had a history of at least one first trimester D&C (> 90% were terminations). The control group of 1,781,786 women included 24,977 with a history of medical management of miscarriage/abortion and 1189 with a history of spontaneous abortion. The primary outcome was preterm birth in a pregnancy subsequent to a history of curettage.
Compared to women with no history of D&C, those with a history of suction curettage had an increased odds ratio (OR) for preterm birth < 37 weeks (OR, 1.29; 95% confidence interval [CI], 1.17-1.42). The association strengthened for very preterm birth: < 32 weeks (OR, 1.69; 95% CI, 1.20-2.38) and < 28 weeks (OR, 1.68; 95% CI, 1.47-1.92). A “dose response” was also observed for women with a history of multiple D&Cs compared with those with no D&C (OR for preterm birth < 37 weeks, 1.74; 95% CI, 1.10-2.76). However, when the control group included only those women with a history of medically managed miscarriage or induced abortion, the association weakened (OR, 1.19; 95% CI, 1.10-1.28).
The authors concluded that the increased risk in association with multiple D&Cs indicates a causal relationship, despite that fact that confounding cannot be excluded. They suggested medical management should be considered for termination of pregnancy and management of miscarriage.
COMMENTARY
No randomized trials will ever be done to assess the outcome of prematurity after surgical abortion. However, this does not mean that we should accept the conclusion of a flawed meta-analysis. The use of systematic review and meta-analysis can boost our confidence when only small studies with limited power suggest an important association. Given that preterm labor represents an important and costly public health problem, strategies that could reduce the risk should be investigated. However, meta-analysis can also amplify misleading information when studies with similar biases are assessed together. In the words of David Grimes and Ken Schultz, “The validity of case-control studies depends on selection of appropriate control groups. Choosing controls might seem deceptively simple, but it can be treacherous.”1 The conclusions of Lemmers et al that suction curettage is associated with an increased risk of premature delivery in a subsequent pregnancy should be carefully evaluated with this in mind.
Given that women tend to under-report a history of abortion, the weak association observed by the authors is likely the result of bias. Numerous studies have documented a substantial degree of under-reporting on self-reported health exposures and outcomes for sensitive topics in sexual and reproductive health, including abortion.2,3 This under-reporting is an important source of bias in studies that attempt to evaluate risk factors for poor obstetrical outcomes. It is not difficult to imagine why this occurs. Women interviewed following a bad outcome (in this case preterm birth) are more likely to recall and report a variety of exposures (including history of prior abortion) than women with a normal birth outcome who have no reason to recall events they would prefer not to disclose. This same relationship confounds research into abortion and breast cancer.4 To be fair, some of the papers included in the Lemmers meta-analysis did use national databases where we would expect accurate reporting of abortion. However, the fact that the meta-analysis included a sample of > 1.8 million women with only 66,000 (approximately 4%) reporting D&C for termination of pregnancy suggests substantial under-reporting overall.
Weak associations (risk ratios < 2) deserve great scrutiny because this is the range in which observed associations are frequently the result of bias and not causality. The purpose of meta-analysis is to increase statistical power to reduce the risk of chance being an explanation for the presence or absence of an observed effect. However, meta-analysis cannot correct observed associations that are the result of confounding bias.5 For example, women who undergo abortion in many regions have baseline characteristics (e.g., smoking, poverty, black race) that are also risk factors for premature birth.6 Failure to adjust for these and other factors (e.g., inter-pregnancy interval) could easily lead to a spurious association.
The appropriate control group consists of women with a history of medical abortion or medical management of miscarriage. In the Lemmers paper, the association weakened (OR, 1.12) when medically managed women were used as controls. Männistö et al evaluated data from the Finnish Register of Induced Abortions and the Medical Birth and the Hospital Discharge Registries.7 Using this national sample, they identified 8294 primigravid women with a history of first trimester termination of pregnancy (3441 medical [MAB], 4853 surgical [SAB]) who had a subsequent pregnancy that resulted in singleton delivery. They found no statistically significant differences in the incidences of preterm birth (4.0% MAB, 4.9% SAB), low birthweight (3.4% MAB, 4.0% SAB), small for gestational age (SGA) infants (2.6% MAB, 2.9% SAB), or placental complications (2.6% MAB, 2.8% SAB) between the two groups. After adjusting for potential confounders associated with adverse pregnancy outcomes (gestational age at the time of termination, inter-pregnancy interval, maternal age, cohabitation status, socioeconomic status, residence, and smoking during pregnancy), MAB was not associated with significantly decreased risks of preterm birth, low birthweight, SGA infant, or placental complications compared to SAB.
A second large study reported data on pregnancy outcomes from 120,033 women with a history of induced abortion collected from the Scottish Morbidity Record.8 No difference was seen in the risk of prematurity in a subsequent pregnancy when women with induced abortion were compared to women with a history of miscarriage. Although a small and marginally statistically significant increase in the adjusted risk of prematurity (< 37 weeks) was found with a history of surgical compared to medical abortion (relative risk [RR], 1.25; 95% CI, 1.07-1.45), this disappeared when the analysis was restricted to very preterm (< 32 weeks [RR, 1.13; 95% CI, 0.81-1.58] or < 28 weeks [RR, 1.38; 95% CI, 0.73-2.61]) births.
The take-home message is that risk of premature delivery in a subsequent pregnancy need not be a counseling point when discussing options for management of unintended pregnancy and pregnancy loss. There are many reasons a woman will choose medical or surgical management. Removing confusing and irrelevant information out of this counseling can better inform the discussion.
REFERENCES
- Grimes DA, Schulz KF. Compared to what? Finding controls for case-control studies. Lancet 2005;365:1429-1433.
- Schroder KE, et al. Methodological challenges in research on sexual risk behavior: II. Accuracy of self-reports. Ann Behav Med 2003;26:104-223.
- Jones EF, Forrest JD. Underreporting of abortion in surveys of U.S. women: 1976 to 1988. Demography 1992;29:113-126.
- Beral V, et al; Collaborative Group on Hormonal Factors in Breast C. Breast cancer and abortion: Collaborative reanalysis of data from 53 epidemiological studies, including 83,000 women with breast cancer from 16 countries. Lancet 2004;363:1007-1016.
- Grimes DA, Schulz KF. Bias and causal associations in observational research. Lancet 2002;359:248-252.
- Bastek JA, et al. Clinical prediction rules for preterm birth in patients presenting with preterm labor. Obstet Gynecol 2012;119:1119-1128.
- Mannisto J, et al. Medical versus surgical termination of pregnancy in primigravid women — is the next delivery differently at risk? A population-based register study. BJOG 2013;120:331-337.
- Bhattacharya S, et al. Reproductive outcomes following induced abortion: A national register-based cohort study in Scotland. BMJ Open 2012;2; pii: e000911. doi: 10.1136/bmjopen-2012-000911.
In a meta-analysis, a history of dilation and curettage for management of miscarriage or termination of pregnancy was associated with an increased risk of preterm birth in a subsequent pregnancy but the association is weak and most likely explained by confounding.
You have reached your article limit for the month. Subscribe now to access this article plus other member-only content.
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content