In previous epidemiologic studies, the prevalence of leiomyomas among pregnant women ranged from 0.1–3.9%.1–5 Four studies reported greater proportions of certain pregnancy, labor, and delivery complications among women with uterine leiomyomas, compared with women without them, including threatened abortion, threatened preterm delivery, abruptio placentae, pelvic pain, premature rupture of membranes (PROM), and breech presentation.1–3,6 Findings on specific complications have been inconsistent. Five studies also examined the risk of cesarean delivery. Four reported higher risk among women with uterine leiomyomas1–3,7 and one found no increase.2
Most studies of pregnancy outcomes in uterine leiomyoma patients had small samples or were composed of participants from only one hospital or clinic.2,3,6,7 Other studies lacked a comparison group of women without uterine leiomyomas7 or did not adjust their findings for possible confounding factors.1,3,7 A literature search using the key words “leiomyoma” or “fibroid” or “uterine myoma” and “pregnancy” or “pregnancy complication” or “pregnancy outcome” or “labor complications” or “delivery” in MEDLINE 1991–November 1999, week 2, identified no publications that used population-based data on this topic.
Materials and Methods
We conducted a retrospective population-based study using data from Washington state birth certificates linked to hospital discharge records for 1987–1993. A total of 2065 women who had delivered singleton live births met the criteria for uterine leiomyoma based on hospital discharge record codes (International Classification of Disease 9 [ICD9]), which included 218.0-.2, 218.9, 654.1654.10-.14. From the remaining records of singleton live births, a group of comparison women without leiomyoma diagnoses was selected at random and frequency matched by infant birth year in a 2:1 ratio to women with leiomyomas.
Information about specific complications in pregnancy, labor, and delivery was collected from birth certificates. Complications of pregnancy recorded on the birth certificate included first-trimester bleeding, placenta previa, abruptio placentae, oligohydramnios, polyhydramnios, anemia, preeclampsia, and PROM. Complications of labor and delivery recorded on the birth certificate included dysfunctional labor, prolonged labor, excessive bleeding, breech presentation, precipitous labor, and cesarean delivery. Neonatal outcomes included 5-minute Apgar scores (0–7, 7–10), infant death (yes, no), birth weight (under 2500 g, 2500–4000 g, and above 4000 g), gestational age (under 38 weeks, 38–40 weeks, and over 40 weeks), and malformations (yes, no). Women with unknown or missing data were excluded from analysis when such data were relevant. For certain variables, data were available only for 1989–1993, and in those instances, analysis was restricted to those years. Seventy-nine women who had more than one birth during the observation period were excluded from analysis.
To assess accuracy of birth certificate cesarean delivery, we compared the odds ratio (OR) using two different methods of coding cesarean delivery. The first coded cesarean delivery if birth certificates indicated it. The second coded a cesarean delivery if the birth certificate or hospital discharge record indicated it. The second method has been reported to be more consistent with medical record information than birth records alone.8
To assess accuracy of hospital discharge reporting of uterine leiomyomas, we reviewed medical records for a sample of births included in this study. One hundred fifty charts were selected at random from one hospital in the Seattle area. Seventy-five were from birth records that indicated a discharge diagnosis of uterine leiomyoma and 75 were from birth records with no discharge diagnosis. Information abstracted from medical charts included uterine leiomyomas (yes, no), size and location of leiomyomas, date of detection, medical procedure used for detection (clinical examination only, ultrasound only, clinical examination and confirmatory ultrasound, or surgical delivery), complications in pregnancy, labor and delivery, gestational age, and 5-minute Apgar scores. Twenty-two records could not be located, and two stored on microfiche were unreadable. Data from the remaining 126 charts were used in analysis.
We used the abstracted information to assess statistical effect of misclassification of women with and without uterine leiomyomas in discharge records. Using a formula provided by Armstrong et al,9 misclassification error was computed for pregnancy and delivery complications.
Unconditional logistic regression was our primary method of analysis. The OR served as the measure of association and 95% CIs were calculated to evaluate the precision of the ORs. Characteristics considered as potential confounders or effect modifiers included maternal age (under 25, 25–34, 35 or more years), race (white, nonwhite), ethnicity (Hispanic, non-Hispanic), marital status (married, not married), residence (rural, urban), smoking status (yes, no), parity (nulliparous, primiparous, or multiparous), prior cesarean delivery (yes, no), maternal weight gain (under 25 lb, 25–40 lb, over 40 lb), hypertension (yes, no), and diabetes (yes, no). We classified women as having established indications for cesarean delivery if birth certificates indicated “fetal distress,” dysfunctional labor, prolonged labor, breech presentation, cephalopelvic disproportion, active genital herpes, or prior cesarean delivery. All potential confounding variables were entered one at a time in a model that contained only the predictor variable of interest. Only variables that appreciably changed the risk were included in the final logistic regression analyses. All analyses were done with Intercooled STATA 5.0 software for Windows (STATA Corp., College Station, TX).10
The prevalence of diagnosed uterine leiomyoma in our study was 0.37%. Women with uterine leiomyomas were similar to women without them in number of pregnancies, previous fetal losses, and maternal weight gain (Table 1). Cases more frequently were over age 35, black, not current smokers, had previous cesareans, or had begun prenatal care in the first trimester of pregnancy.
Women with leiomyomas were twice as likely as women without to have had a complication during pregnancy (multivariate OR 1.87, 95% CI 1.59, 2.20) (Table 2). When we examined the association between uterine leiomyoma and specific pregnancy complications, we found cases were more likely to have had bleeding during the first trimester, abruptio placentae, and PROM.
Women with uterine leiomyomas at discharge were approximately twice as likely to have complications during labor or delivery than women without them (multivariate OR 1.90, 95% CI 1.65, 2.18) (Table 3). Dysfunctional labor, prolonged labor, breech presentation, and cesarean delivery strongly associated with the presence of uterine leiomyomas. After stratifying the population by whether other indications for cesarean delivery were noted on birth certificates, the OR was higher for women without other indications for cesarean (multivariate OR 7.59, 95% CI 5.47, 10.53), compared with women with other indications (multivariate OR 5.26, 95% CI 3.98, 6.95) (Table 4).
Women with uterine leiomyomas were at higher risk for poor birth outcomes (Table 5). When we examined the association between uterine leiomyomas and specific birth outcomes, we observed that cases were more likely than controls to have newborns with 5-minute Apgar scores of less than 7, with low birth weight (LBW; less than 2500 g), or with malformations.
When birth records and hospital discharge summaries were used to code women with cesareans, 433 additional births by cesarean (6.9%) were identified. The multivariate-adjusted OR for the association between uterine leiomyomas and cesarean delivery, using the method that coded cesarean if the birth records or the hospital discharge reports indicated cesarean, was calculated at 10.43. The corresponding estimate for analyses based only on birth certificate coding of cesarean was 6.39.
Of 126 medical records reviewed, 60 contained diagnoses of uterine leiomyomas and 66 had no diagnoses. Of 60 women with leiomyomas noted on computerized hospital discharge, 57 (95.0%) were confirmed to have leiomyomas on medical record review. Of 66 women without leiomyoma diagnoses at discharge, the medical records for 58 (87.9%) did not mention uterine leiomyomas. For nearly half of the sample women with uterine leiomyomas, the first mention of uterine leiomyoma in the medical record was in the delivery report.
In the sample of deliveries for which we reviewed medical records, cesarean delivery varied markedly among women with and without leiomyomas. Among women with leiomyomas, 66.6% delivered by cesarean. Among women with uterine leiomyomas who delivered by cesarean, the diagnosis for 72.5% was first noted on the delivery report, compared with 5.0% who delivered vaginally. Thirty-eight percent of women with uterine leiomyomas and 25% without, who had cesareans, had the procedure scheduled.
The correlation of leiomyoma status on computerized hospital discharge files and medical records was 0.73. Using the correlation, we estimated a corrected OR for the association between leiomyoma and pregnancy and delivery complications. The corrected OR for the association of leiomyoma with prolonged labor was 1.54, 32% higher than the OR of 1.17. The corrected ORs for other outcomes were 3.08 for first-trimester bleeding, 2.98 for PROM, 5.54 for low Apgar scores, 3.64 for LBW, 3.17 for dysfunctional labor, and 13.36 for breech presentation.
Complications of pregnancy, labor, and delivery were nearly twice as frequent among women with diagnosed uterine leiomyomas than those without them. The proportion of women who had cesareans was higher among women with uterine leiomyomas compared with women without them, particularly when there was no other indication for surgical delivery. Infants of women with uterine leiomyomas were of lower birth weight and had lower 5-minute Apgar scores than infants of women without them.
An important limitation of this study was the method by which we determined whether women had or did not have uterine leiomyomas. The lower prevalence of pregnant women with uterine leiomyomas in our study compared with previous studies might suggest that hospital discharge records underreport leiomyomas. Underreporting might have occurred if a woman did not have a sonographic or physical examination, leiomyoma was undetected during the examination, or the diagnosis was not reported on the hospital discharge record.
Various factors might explain why prevalence of uterine leiomyomas in our population-based setting might have been lower than reports from clinic or hospital-based studies. Individual clinics or hospitals might have only women who had ultrasound examinations or were at high risk. The prevalence of uterine leiomyomas increases with age, so differences in age distribution between study populations could lead to differences between our study and previous research. If leiomyoma size is an important determinant of pregnancy-related complications, then we would expect only a slight attenuation of associations when some women with uterine leiomyomas were misclassified.
It is also possible that the ascertainment of uterine leiomyomas differed according to pregnancy and delivery outcomes of interest. For example, our finding from medical record review that women with leiomyomas who had cesareans were much more likely to have had their tumors diagnosed at delivery likely indicates that operations led to finding tumors. To the extent that that occurred in all hospital discharges, our estimates of association between uterine leiomyomas and cesarean would be biased upward. If certain complications were more likely to be reported on the hospital discharge summary for women who had leiomyomas than women without them, the OR estimates for those complications might be artifactually elevated.
In our review of medical charts in a sample of women with and without uterine leiomyomas, recorded in the hospital discharge data, we observed that 10% of women had uterine leiomyoma diagnoses not noted on discharge. The misclassification of these women would have resulted in an underestimate of the association between uterine leiomyoma and birth outcomes reported in the present study. The magnitude of the effect of potential misclassification was estimated in the present study.
The advantages of our study include the availability of information about neonatal characteristics, the availability of information about confounding factors, and a population-based sample. The comparison of hospital discharge records with medical chart reports allowed us to assess validity of our data.
The results of this study were consistent with some previous research about the relationship between uterine leiomyomas and pregnancy, labor, and delivery complications. Several studies that addressed breech presentation among women with uterine leiomyomas reported statistically significant positive associations, and some have suggested that size and location of the leiomyoma might predict the magnitude of the risk.1,6 Our fourfold increased proportion of breech presentation among women with uterine leiomyomas relative to controls was consistent with this research. Our finding of a sixfold increased proportion of cesareans among women with uterine leiomyomas, relative to controls, was consistent with previous findings. For example, Katz et al reported the proportion of cesarean delivery at 34% with leiomyomas, compared with an overall hospital incidence of 22%.3 Veragani et al found cesareans in 23.4% of leiomyoma patients, compared with 12.1% in the general population.7 The three studies that used control groups without leiomyomas found the proportion of cesarean delivery to be 23.4% versus 12.1%, 20.5% versus 19.3%, and 40% versus 33%, respectively.1,2,6 We found greater disparity between groups in proportion of cesareans than previously (multivariate OR 6.39, 95% CI 5.46, 7.50). Some of that difference is likely caused by biased ascertainment of uterine leiomyomas at surgery. The near fourfold increased proportion of abruptio placentae among women with uterine leiomyoma in our study suggests a stronger association than previously reported.1,2,6 Our finding of an increased proportion of PROM among women with uterine leiomyoma is consistent with one study and inconsistent with two others.1,2,6
The biologic basis for the association between pregnancy, labor, or delivery complications and uterine leiomyoma is unclear. Some research suggested that leiomyomas that are behind the placenta or in the lower uterine segment increase the likelihood of delivery complications.1,6,7 Uterine leiomyomas also might decrease uterine distensibility, or present mechanical obstructions that restrict space, limit fetal movement, or lessen the force of contractions.1,7 The reduced risk of precipitous labor among women with uterine leiomyomas lends support to the hypothesis that uterine leiomyomas interfere with uterine contractions.
In this large population-based study, we observed that uterine leiomyoma was strongly related to breech presentation, dysfunctional labor, cesarean delivery, LBW, low gestational age, and low Apgar scores. Women with uterine leiomyoma were less likely to have precipitous labor. These results were consistent with hypotheses that uterine leiomyomas might interfere with normal labor and delivery by mechanical processes. Future investigations should focus on how to minimize risks among pregnant women with uterine leiomyomas.
1. Rice JP, Kay HH, Mahony BS. The clinical significance of uterine leiomyomas in pregnancy. Am J Obstet Gynecol 1989;160:1212–6.
2. Exacoustos C, Rosati P. Ultrasound diagnosis of uterine myomas and complications in pregnancy. Obstet Gynecol 1993;82:97–101.
3. Katz VL, Dotters DJ, Droegemueller W. Complications of uterine leiomyomas in pregnancy. Obstet Gynecol 1989;73:593–6.
4. Burton CA, Grimes DA, March CM. Surgical management of leiomyomata during pregnancy. Obstet Gynecol 1989;74:707–9.
5. Hasan F, Arumugam K, Sivanesaratnam V. Uterine leiomyomata in pregnancy. Int J Gynaecol Obstet 1991;34:45–8.
6. Davis JL, Ray-Mazumder S, Hobel CJ, Baley K, Sassoon D. Uterine leiomyomas in pregnancy: A prospective study. Obstet Gynecol 1990;75:41–4.
7. Vergani P, Ghidini A, Strobelt N, Roncaglia N, Locatelli A, Lapinski RH, et al. Do uterine leiomyomas influence pregnancy outcome? Am J Perinatol 1994;11:356–8.
8. Parrish KM, Holt VL, Connell FA, Williams B, LoGerfo JP. Variations in the accuracy of obstetric procedures and diagnoses on birth records in Washington State, 1989. Am J Epidemiol 1993;138:119–27.
9. Armstrong B, White E, Saracci R. Principles of exposure measurement in epidemiology. In: Kelsey J, Marmot M, Stolley P, Vessey M, eds. Monographs on epidemiology and biostatistics. Vol. 21. New York: Oxford University Press, 1994:1–351.
10. Stata Corporation. Stata statistical software: Release 5.0. College Station, TX: Stata Corporation, 1997.
© 2000 The American College of Obstetricians and Gynecologists
This article has been cited