Spontaneous abortion is one of the most common complications of pregnancy. It is estimated that as many as 12–15% of clinically recognized pregnancies, and as many as 17%-22% of all pregnancies, result in spontaneous abortion.1,2 The diagnosis of spontaneous abortion currently depends on diagnostic strategies involving a combination of ultrasonography and serum human chorionic gonadotropin (hCG) levels. When miscarriage is not definitively diagnosed with ultrasonography, diagnosis requires repetitive testing over the course of several days to weeks. This process not only delays treatment, but can be emotionally trying for patients. Identifying women at high risk for miscarriage before or during the earliest stages of pregnancy might allow clinicians to treat and counsel high risk patients appropriately to minimize the psychological and physical risks associated with this condition.
Although the most common cause for miscarriage is the presence of a genetic abnormality in the fetus, accounting for 54–76% of cases, a large proportion of spontaneous abortions remain unexplained.3 Previous studies have shown that various historical factors are related to miscarriage such as increased maternal age, ethnic origin, past or current use of intrauterine device (IUD), prior sexually transmitted diseases, artificial reproductive technologies, gravidity, parity, and prior fetal losses.4,5 Other studies have identified associations between miscarriage and clinical factors such as hCG levels and ultrasound findings.3,6 Infectious agents such as bacterial vaginosis and Chlamydia have also been linked to spontaneous abortion.7–9 However, these studies are limited, particularly because of poor control group selection. For instance, results from studies comparing healthy intrauterine pregnancies to spontaneous abortions may be biased, because the cases and controls are not taken from the same population and therefore are not comparable. There is evidence that early complicated pregnancies (spontaneous abortion, ectopic pregnancy, and ongoing intrauterine pregnancy with symptoms) have similar risk factors.10 Therefore, selecting cases of miscarriage and controls from a population with early complicated pregnancy may avoid selection bias.
The aim of this study was to evaluate the association between miscarriage and various clinical symptoms and historical factors in a cohort of women presenting with pain or bleeding in the first trimester of pregnancy in women whose diagnosis could not be made at presentation. This population represents approximately 20% of women who present with a symptomatic first trimester pregnancy.11,12
MATERIALS AND METHODS
Women presenting to the University of Pennsylvania Medical Center with pain or bleeding or both in pregnancy, before definitive diagnosis, during the years January 1, 1990, to July 31, 1999, were included in this study. Women were evaluated with a history, physical examination including pelvic examination, pelvic ultrasonography, cervical cultures, and an hCG level. Women whose diagnosis was unclear upon presentation were included in a database at the University of Pennsylvania. All such patients had ultrasound findings that revealed no evidence of an intrauterine or extrauterine gestation in the face of an hCG value below the established discriminatory zone. Patients were followed up on an outpatient basis until a definitive diagnosis of ongoing viable intrauterine pregnancy, ectopic pregnancy, or miscarriage was made. In general, patients were followed up closely with serial hCG measurements and pelvic ultrasonography. In some cases, surgery in the form of laparoscopy or dilation and curettage (D&C) was required to make a definitive diagnosis. Approval to conduct the study was obtained from the Institutional Review Board of the University of Pennsylvania.
Data were entered prospectively into the computerized database by clinical staff caring for the patient. Potential predictors derived from historical factors were race, age, gravidity, parity, number of live births, number of previous ectopic pregnancies, number of previous spontaneous abortions, number of previous induced abortions, number of cesarean deliveries, history of pelvic inflammatory disease (PID, defined as requiring inpatient treatment), history of cervical infection with gonorrhea or Chlamydia (outpatient treatment), IUD use, and history of pelvic surgery (excluding prior cesarean delivery and tubal surgery). Clinical signs and symptoms at presentation analyzed as predictors of spontaneous abortion were presence of pain, severity of bleeding, hCG level, ultrasound examination, and current infection with gonorrhea or Chlamydia detected by cervical testing.
Bleeding was categorized according the patient’s subjective assessment at presentation. The categories were none, mild, moderate, or severe, defined as no bleeding, less than, equal to, or more than regular menstrual bleeding respectively. The presence of pain was also reported by the patient. This variable was categorized as yes or no. “Yes” was defined as a chief complaint of pain. “No” was defined as a chief complaint of other than pain (ie, bleeding). Ultrasonography at presentation results were categorized as probable intrauterine pregnancy (possible intrauterine sac without yoke sac or fetal pole), nondiagnostic, or probable ectopic pregnancy (possible extrauterine sac or mass uterine sac without yoke sac or fetal pole).
The final pregnancy outcome was confirmed as spontaneous abortion by either a histopathology of products of conception from suction D&C or by an hCG level of 5 IU/mL or less with no intervention. The presence of a viable, ongoing normal intrauterine pregnancy was confirmed by interval growth of an intrauterine gestation with visualization of an intrauterine yolk sac, fetal pole, or heartbeat on ultrasonography. The diagnosis of ectopic pregnancy was confirmed by either visualizing an extrauterine gestational sac or an increase in serum hCG after no products of conception were noted after D&C.
Cases were defined as women ultimately definitively diagnosed with a spontaneous abortion. Controls were defined in 2 ways, and 2 separate analyses were performed. Our primary analysis defined controls as women presenting with similar symptoms but not diagnosed with a miscarriage. Thus, the control group included women with either an ectopic pregnancy or an ongoing intrauterine pregnancy. The second analysis used the definition of controls as women presenting with symptoms but diagnosed with an ongoing intrauterine pregnancy (women diagnosed with ectopic pregnancies were excluded).
This was a case–control study. Statistical analysis was performed using SAS 8 (SAS Institute Inc., Cary, NC). Descriptive statistics were initially used to summarize the distribution of variables in the entire population. First, univariate tests were performed to assess the association between various potential risk factors and spontaneous abortion using the Student t test and Pearson’s χ2 for continuous and categorical variables, respectively. To explore differences in univariate associations resulting from classification of information, some variables (age, parity, history of surgery) were analyzed as both continuous and categorical variables.
Next, stratified analyses were performed to evaluate the potential for confounding and effect modification. Historical and clinical variables were first examined individually to check for homogeneity of the odds ratios when stratified by age. This was done by evaluating stratified 2 × 2 tables and testing for interactions between the risk factor of interest and age within a logistic regression model. When heterogeneity was not present, these variables were considered in the logistic regression models. For purposes of analysis of categorical values, reference categories included age 25–29 years, hCG less than 500 IU/mL, and the presence of a nondiagnostic ultrasonogram. A multivariate logistic regression model was then constructed using manual selection of confounding variables and backward stepwise selection of variables. At each step, the variable with the largest P value was removed from the model and this process was repeated until all the variables had a P = .05. If removal of a variable affected the coefficient estimates of the other variables by 15% or more, that variable was considered a confounder and retained in the model.13
For this study, we compared women with spontaneous abortion with those with ectopic pregnancy and ongoing, viable intrauterine pregnancies. We also created a model by excluding all cases of ectopic pregnancy in order to distinguish risk factors for spontaneous abortion compared with ongoing, viable intrauterine pregnancies.
A total of 2,352 patients who presented with pain or bleeding in the early first trimester of pregnancy and whose diagnoses could not be made at presentation were evaluated from January 1, 1990, to July 31, 1999. Of these patients, 30 patients were diagnosed with a molar pregnancy, and 296 patients were lost to follow-up. These women were excluded from the database, leaving a total of 2,026 patients. Of these, 1,192 women ultimately definitively diagnosed with a spontaneous abortion were selected as cases; 834 women diagnosed with either an ectopic pregnancy (n = 367) or an intrauterine pregnancy (n = 467) served as the control population.
There were several unadjusted risk factors for spontaneous abortion. Study of racial distribution showed that approximately 94.5% of the population was African-American. Race was not significantly associated with spontaneous abortion. Overall, women with spontaneous abortion were younger than controls. When age was explored as a categorical variable, it seemed that women diagnosed with miscarriage were more likely to be either less than 25 or at least 35 years of age (Table 1). No historical factor assessed was associated with a statistically significant increased risk of spontaneous abortion. However, several historical factors were associated with a decreased risk of spontaneous abortion, including prior live birth, prior ectopic pregnancy, prior spontaneous abortion, prior pelvic surgery, history of PID, or history of a cervical infection treated as an outpatient (Table 2. Clinical factors at presentation associated with an increased risk of spontaneous abortion include moderate to severe bleeding and nondiagnostic ultrasonogram. Clinical factors at presentation associated with a decreased risk of spontaneous abortion include pain and an hCG level greater than 500 IU/mL (Table 3).
Final, adjusted risk factors for spontaneous abortion compared with all controls are shown in Table 4
The final multivariate model included the following variables: age, bleeding at presentation, pain at presentation, current infection with gonorrhea or Chlamydia, and hCG level at presentation. Variables associated with increased risk of miscarriage included age younger than 25 years and moderate or severe bleeding at the time of presentation. Factors associated with a decreased risk of spontaneous abortion included pain, a current infection with gonorrhea or Chlamydia, or an hCG value 500 IU/mL or more at presentation. In the final analysis, no statistically significant association was found between miscarriage and gravidity, parity, prior ectopic pregnancy, prior spontaneous abortion, prior elective termination of pregnancy, prior pelvic surgery, prior cesarean delivery prior PID, prior gonorrhea or Chlamydia infection treated as an outpatient, results of the ultrasonography at presentation, or prior IUD use.
Final, adjusted risk factors for spontaneous abortion compared only with viable ongoing intrauterine pregnancies are shown in Table 5.
When spontaneous abortion was compared with intrauterine pregnancy only, excluding cases of ectopic pregnancy, we found that it was even more strongly associated with moderate to severe bleeding at presentation than in the previous analysis. Again, hCG level more than 500 IU/mL was associated with a decreased risk of spontaneous abortion. Age was marginally associated with spontaneous abortion, with more women aged at least 35 years experiencing spontaneous abortion compared with controls. History of spontaneous abortion was marginally associated with spontaneous abortion. In the final analysis, no statistically significant association was found between miscarriage and gravidity, parity, prior ectopic pregnancy, prior elective termination of pregnancy, prior pelvic surgery, prior cesarean delivery, prior PID, prior infection with gonorrhea or Chlamydia treated as outpatient, pain at presentation, cervical infection at presentation, nondiagnostic ultrasonogram at presentation, or prior IUD use.
The main advantage of this study is that the ascertainment of risk factors was performed in a manner with less bias than previous observational studies evaluating the risk factors for miscarriage. Selection bias has been minimized by including cases and controls from the same underlying population of patients. Additionally, all clinical data were collected prospectively before ascertainment of definitive diagnosis, which helps to eliminate recall bias. Additionally, we focused our study on women who represent a clinical conundrum to clinicians. Moreover, this study combined historical risk factors with factors regarding the patients’ presentation. We feel that all of these factors make this study less prone to bias and more pertinent to a clinically important population of women.
We found that in general age is an independent risk factor for miscarriage. Women between the ages of 25 and 30 years had the lowest risk of spontaneous abortion, whereas those with the highest risk of spontaneous abortion were at the extremes of age. Specifically, women with miscarriage were twice as likely to be aged younger than 20 years compared with those without miscarriage. If analysis was restricted to the comparison of women with an ongoing intrauterine pregnancy, the association of age followed a similar trend but was weaker. These findings are consistent with reports demonstrating a higher risk of spontaneous abortion in teenage pregnancies.14 In the analysis excluding women with ectopic pregnancies, it seemed that women aged 35 years and older also had double the risk of spontaneous abortion. Other studies support an increased risk of spontaneous abortion with advanced maternal age as well.15–20
As expected, moderate to severe bleeding was strongly associated with the diagnosis of miscarriage. Specifically, women who eventually miscarried were more than 18 times as likely to experience bleeding compared with women with an ongoing, viable intrauterine pregnancy. According to other studies, 92% of miscarriages are accompanied by vaginal bleeding, and bleeding or passage of vaginal clots similar to menses is a significant clinical predictor of spontaneous abortion.21,22
Conversely, the presence of pain at presentation was negatively associated with miscarriage. We found that if women presented with a chief complaint of pain they were more likely to have something other than a miscarriage. The reason for this association may be that pain is more likely associated with risk of ectopic pregnancy. When controls were restricted to intrauterine pregnancy only (and women with ectopic pregnancy were excluded) the negative association of pain with miscarriage was no longer apparent.
We also found that women with miscarriage were less likely to be diagnosed with gonorrhea or Chlamydia on presentation. This finding should not be interpreted to mean that cervical infection is protective of miscarriage. On the contrary, we believe that discharge or bleeding related to the infection, and not an impending miscarriage, prompted evaluation and that most of these women had viable, ongoing pregnancies.
Women with miscarriage were roughly one half as likely to have an hCG value 500 mIU/mL or more compared with women diagnosed with a viable, ongoing pregnancy or ectopic pregnancy. In other words, women with an hCG level greater than 500 IU/mL were more likely to have an ongoing viable intrauterine pregnancy or ectopic pregnancy. We believe that it makes intuitive sense that hCG levels are likely to be low in nonviable pregnancies that may be spontaneously resolving. On the other hand, it makes sense that pregnancies that are continuing to grow (such as intrauterine pregnancy or ectopic pregnancy) have higher hCG values. Although there is good data showing that hCG is the most valuable diagnostic test before 6 weeks gestation, particularly in the presence of a nondiagnostic ultrasonogram, 1 value in isolation cannot be used reliably to distinguish a nonviable pregnancy from an ectopic pregnancy or viable intrauterine pregnancy.23
The following factors were not associated with spontaneous abortion in this study: gravidity, parity, prior ectopic pregnancy, prior spontaneous abortion, prior elective termination of pregnancy, prior pelvic surgery, prior cesarean delivery, prior PID, prior infection with gonorrhea or Chlamydia treated as an outpatient, or prior IUD use. Although there is some evidence from the literature to support the association between some of these factors and miscarriage, we were not able to demonstrate statistically significant associations in this study.24–25 In particular, several studies have found that previous miscarriage is a strong risk factor for spontaneous abortion.4,5 However, the risk seems to be negligible with a history of only 1 prior miscarriage. In this study, there were few women with a history of recurrent miscarriage, so our finding seems to be consistent with the existing literature. In addition, reports are conflicting with respect to the risk of miscarriage after elective termination of pregnancy. Although studies have reported a risk 2.3 to 8.1 times higher after 2 to 3 or more abortions,24,25 our results corroborate another study that found no association of previous induced abortion with subsequent spontaneous abortion.
Although this study demonstrates that age, the presence of bleeding, pain, history of cervical infection, and hCG levels are independently associated with miscarriage, no single test can be used to accurately distinguish a miscarriage from an ongoing intrauterine pregnancy or ectopic pregnancy. Following up patients with a combination of hCG and ultrasonography remains the optimal diagnostic strategy to evaluate patients with symptomatic early pregnancy.
1. Zinaman MJ, Clegg ED, Brown CC, O’Connor J, Selevan SG. Estimates of human fertility and pregnancy loss. Fertil Steril 1996;65:503–9.
2. Ellish NJ, Saboda K, O’Connor J, Nasca PC, Stanek EJ, Boyle C. A prospective study of early pregnancy loss. Hum Reprod 1996;11:406–12.
3. Fistarol M, Ceriani SB, Borruto F. Incidence of genetic alteration in spontaneous abortion: attempted correlations with betaHCG values and with the ultrasonic image [in Italian]. Minerva Ginecol 1999;51:265–70.
4. Coste J, Job-Spira N, Fernandez H. Risk factors for spontaneous abortion: a case control study in France. Hum Reprod 1991;6:1332–7.
5. Risch HA, Weiss NS, Clarke EA, Miller AB. Risk factors for spontaneous abortion and its recurrence. Am J Epidemiol 1988;128:420–30.
6. Parazzini F, Chatenoud L, Tozzi L, Benzi G, Dal Pino D, Fedele L. Determinants of risk of spontaneous abortions in the first trimester of pregnancy. Epidemiology 1997;8:681–3.
7. von Rechenberg KN. Percentile change in the beta-HCG concentration in maternal serum in disordered early pregnancy [in German]. Z Geburtshilfe Perinatol 1989;193:72–6.
8. Jouppila P, Huhtaniemi I, Tapanainen J. Early pregnancy failure: study by ultrasonic and hormonal methods. Obstet and Gynecol 1980;55:42–7.
9. Simpson JL, Mills JL, Kim H, Holmes LB, Lee J, Metzger B, et al. Infectious processes: an infrequent cause of first trimester spontaneous abortions. Hum Reprod 1996;11:668–72.
10. Bouyer J, Coste J, Shojaei T, Pouly JL, Fernandez H, Gerbaud L, et al. Risk factors for ectopic pregnancy: a comprehensive analysis based on a large case–control, population based study in France. Am J Epidemiol 2003;157:185–94.
11. Barnhart K, Mennuti MT, Benjamin I, Jacobson S, Goodman D, Coutifaris C. Prompt diagnosis of ectopic pregnancy in an emergency department setting. Obstet Gynecol 1994;84:1010–5.
12. Barnhart KT, Simhan H, Kamelle S. Diagnostic accuracy of ultrasound above and below the beta-hCG discriminatory zone. Obstet Gynecol 1999;94:583–7.
13. Maldonado G, Greenland S. Simulation study of confounder-selection strategies. Am J Epidemiol 1993;138:923–36.
14. Shawky S, Milaat W. Early teenage marriage and subsequent pregnancy outcome. Eastern Mediterr Health J 2000;6:46–54.
15. Wilcox A. Surveillance of pregnancy loss in human populations. Am J Ind Med 1983;4:285–91.
16. Saraiya M, Green CA, Berg CJ, Hopkins FW, Koonin LM, Atrash HK. Spontaneous abortion-related deaths among women in the United States—1981–1991. Obstet Gynecol 1999;94:172–6.
17. de la Rochebrochard E, Thonneau P. Paternal age and maternal age are risk factors for miscarriage; results of a multicenter European study. Hum Reprod 2002;17:1649–56.
18. Hakim RB, Gray RH, Zacur H. Infertility and early pregnancy loss. Am J Obstet Gynecol 1995;172:1510–7.
19. Smith KE, Buyalos RP. The profound impact of patient age on pregnancy outcome after early detection of fetal cardiac activity. Fertil Steril 1996;65:35–40.
20. Berry SW, Brambati B, Eskes TK, Exalto N, Fox H, Geraedts JP, et al. The Euro-Team Early Pregnancy (ETEP) protocol for recurrent miscarriage. Hum Reprod 1995;10:1516–20.
21. Baszak E, Sikorski R, Milart P, Wojcik D. Clinical features of missed abortion [in Polish]. Ginekol Pol 2001;72:1069–72.
22. Chung TK, Sahota DS, Lau TK, Mongelli JM, Spencer JA, Haines CJ, et al. Threatened abortion: prediction of viability based on signs and symptoms. Aust N Z J Obstet Gynecol 1999;39:443–7.
23. Gerhard I, Runnebaum B. Predictive value of hormone determinations in the first half of pregnancy. Eur J Obstet Gynecol Reprod Biol 1984;17:1–17.
24. Levin AA, Schoenbaum SC, Monson RR, Stubblefield PG, Ryan KJ. Association of induced abortion with subsequent pregnancy loss. JAMA 1980;243:2495–9.
25. Infante-Rivard C, Gauthier R. Induced abortion as a risk factor for subsequent fetal loss. Epidemiology 1996;7:540–542.