AMERICAN JOURNAL OF PERINATOLOGY/VOLUME 7, NUMBER 4
October 1990
RISK FACTORS FOR PRETERM PREMATURE RUPTURE OF THE FETAL MEMBRANES Carolyn B. Hadley, M.D., Denise M. Main, M.D., and Steven G. Gabbe, M.D.
Preterm premature rupture of the membranes (PPROM) is a significant cause of prematurity, accounting for approximately one third of preterm births in the United States. PPROM occurs in approximately 0.7—2% of all pregnancies nationally, and has a reported recurrence rate of 21%. The elucidation of potential risk factors for PPROM could contribute to a better understanding of its etiology. To study the contributions of 20 potential risk factors, we undertook a case-control study in our clinic population, which has a 5-6% incidence of PPROM. One hundred and thirty-three patients experiencing PPROM were matched for race, age, parity and gestational age with undelivered patients. Studies performed included ultrasonographic examinations, blood levels of ascorbic acid and zinc, microbiologic assays, patient questionnaires, and chart reviews. After stratification of both groups into subgroups based on matching criteria, summary tests of significance and Mantel-Haenszel tests of odds ratios were performed. On univariate analysis the following factors achieved significance at the p < 0.05 level with 95% confidence intervals: 1) previous history of PPROM 2) smoking (dose related) 3) fundal location of the placenta in the present pregnancy. 4) a prior history of cerclage. After regression analysis, we concluded that smoking and history of previous PPROM were found to be risk factors for PPROM in our inner city black population.
The term "premature rupture of the membranes" (PROM) is generally accepted as referring to the leakage of amniotic fluid through the cervical os more than one hour prior to the onset of labor. Preterm PROM (PPROM) includes all cases of PROM occurring before 37 completed weeks gestational age. PPROM occurs in 0.7-2.0% l of all pregnancies, but accounts for approximately 30% of all preterm births2 and is the single most common diagnosis leading to admission to newborn intensive care nurseries. In a population-based study in North Carolina, Meis and his co-workers noted that PPROM is a much more frequent cause of preterm births in indigent patients than in private patients.3 In our clinic population of inner city black women, PPROM accounts for 5-6% of all deliveries, a figure 2-3 times the national average. Given the high incidence of PPROM among our population, we recognized the importance of identifying possible risk factors prospectively.
Many theories have been proposed to explain the etiology of PPROM. Most have focused on the disruption of the mechanical integrity of the membranes. 45 Damage to the membranes may be caused by substances produced by genital tract bacteria themselves or may result from the inflammatory response of the host.67 Decreased maternal levels of ascorbic acid, copper and zinc have been associated with PPROM.89 Smoking is known to impair protein metabolism and to reduce the levels of amino acids, Vitamin B12 and ascorbic acid.10 Meyer and Tonascia reported a three-fold increase of PPROM between 20-34 weeks' gestational age for smokers as opposed to non-smokers.l J Coitus has also been implicated as contributing to the incidence of PPROM12 in one study but not in others 1314 Toth et al implicated preexisting infection of the uterine cavity as a predisposing factor for premature rupture of the membranes and preterm delivery.15 The recurrence risk for PPROM has been noted to be as high as 21%. 16
From the Departments of Obstetrics and Gynecology, Hospital of the University of Pennsylvania, Medical College of Pennsylvania, University of California at San Francisco, and Ohio State University College of Medicine Reprint requests: Dr. Hadley, Dept. of Obstetrics and Gynecology, Medical College of Pennsylvania, 3300 Henry Avenue, Philadelphia, PA 19129 374
Copyright © 1990 by Thieme Medical Publishers, Inc., 381 Park Avenue South, New York, NY 10016. All rights reserved.
Downloaded by: Universite Laval. Copyrighted material.
ABSTRACT
Finally, a number of gynecologic procedures and pregnancy-related conditions have been thought to contribute to the occurrence of PPROM. Bleeding during the early part of the current pregnancy was found by Joffee to predispose patients to PPROM.17 Pregnancies with hydramnios tend to be more susceptible to PPROM, presumably because of mechanical stress on the cervix and membranes. Fundal placental location, which places the weakest point in the membranes over the cervical os, is felt to predispose patients to PPROM.18 Marginal cord insertions have been noted more frequently in patients with PPROM.19 To better elucidate the risk factors associated with PPROM and because of the absence of a controlled study in the literature which addresses most of the potential risk factors for this important perinatal complication, we undertook this case-control study.
MATERIALS AND METHODS
The design of this study, i.e. case control with non-random ascertainment of cases and controls, enabled the authors to work with a smaller study sample size than would have been required for a prospective study. Cases and controls were obtained sequentially. According to Schlesselman, this sample size with matched controls is adequate to test for a relative risk or 2.0 with Type I and II errors of 0.05 and 0.20, respectively.20 For the purposes of this study PPROM refers to rupture of the membranes occurring more than 1 hour prior to the onset of labor at a gestational age of less than 37 weeks. Gestational age was determined by ultrasonographic and clinical criteria. Rupture of membranes was diagnosed by the presence of a vaginal pool, a positive nitrazine test, microscopic ferning and by documenting decreased amniotic fluid on ultrasonographic exam. For one year, each patient on the Clinic Service at the Hospital of the University of Pennsylvania, Philadelphia, who was diagnosed as having PPROM was matched to an undelivered control patient who was being followed in the routine Obstetrical Clinic at the same institution. Undelivered controls were selected because it was felt that these patients were typical of our general obstetrical population. Patients considered to be at high risk by virtue of their medical and obstetrical histories were excluded from the control population. The controls were selected by the senior author by matching each new patient with PPROM to the next available patient in the routine obstetrical clinic whose matching parameters fit the necessary criteria. All study patients were black and had singleton gestations. The study subjects were matched to controls for age (greater or less than 20 yrs) parity (parous vs. nulliparous) and gestational age (within 1 —2 weeks). It was felt that such careful matching would limit confounding factors. The patients were matched for
age because of the potential difference in nutritional and social habits between teenagers and adults. The mean age for patients under 20 years was 17.8 years for the PPROM group and 17.9 years for controls. The mean age for patients over 20 years was 26.8 years for the PPROM group and 26.3 years for controls. Study subjects were matched for parity to take into account the potential contributions of the patient's prior obstetrical history. Patients were matched for gestational age because of the potentially greater influence of intrauterine infection at earlier gestational ages. All patients were under 37 weeks' gestational age. Every effort was made to match gestational ages to within one week with the outside limit being 2 weeks. There were 133 PPROM and 133 control patients recruited into the study which was approved by the Institutional Review Board. Written consent was obtained from each participating patient at the time of the interview. The patient questionnaire was administered as a structured interview by the senior author (CBH) to all participants. After delivery the answers in the questionnaire and medical history were verified by a thorough check of the entire medical record. Items covered in the questionnaire and medical record review are listed in Table 1. Each study patient underwent an ultrasonographic examination to determine placental location. In addition, cervical cultures for Neisseria gonorrhoeae, and a vaginal culture for Group B streptococcus were obtained in all patients. Serum ascorbic acid levels and plasma zinc levels were obtained in smaller groups of patients and their matched controls. The number of patients tested for zinc and ascorbic acid levels was limited by the cost of these assays; however, all matching criteria were fulfilled. Both specialized assays were performed in the laboratories of Smith-Kline Bioscience, Inc. (King of Prussia, PA). Cut-off values were provided by the laboratories and were as follows: ascorbic acid, 0.2 mg/dl and plasma zinc, 60 mcg/dl.
Statistical Methods
The univariate analyses of the effect of each potential risk factor took into account the matching scheme. Specifically, a cross classification of the matching variables produced strata in which the cases and controls were homogeneous with respect to the matching variables. The risk of PPROM was estimated within each strata; summary tests of significance and odds ratios, with their 95% confidence intervals, were obtained using the Mantel-Haenszel method. In addition to determining whether the presence of smoking was a risk factor, the authors tested for a linear dose response relationship using a chi-square test for linear trend. Multiple logistic regression was used to determine which factors contributed independently at least to some extent to elevating the risk of PPROM after controlling for other factors which were held 375
Downloaded by: Universite Laval. Copyrighted material.
PRETERM PREMATURE RUPTURE/Hadley, Main, Gabbe
AMERICAN JOURNAL OF PERINATOLOGY/VOLUME 7, NUMBER 4
October 1990
PPROM (%) N = 133 Historical factors Dilatation and curettage First trimester spontaneous abortions First trimester therapeutic abortions Second trimester spontaneous abortions Second trimester therapeutic abortions Previous preterm delivery without PPROM Previous PPROM Cere I age Cone biopsy Pelvic inflammatory disease requiring hospitalization Uterine malformations (by hysterosalpingogram) DES exposure (by history and exam) Current pregnancy Smoking (number of cigarettes per day) Coitus within 1 week Bleeding after twelve weeks No prenatal care Cerclage Hydramnios (pocket > 8 cm by ultrasound) Fundal placental location Microbial infection/colonization Group B streptococcus Neisseria gonorrhoeae
Control (%) N = 133
12.8 18.8
16.5
37.6
36.8
8.3
3.0
6.8
3.8
4.5
5.3
33.1
12.8
5.3 1.5 1.5
0.8 0.0 5.2
2.3
2.3
0.8
0.0
55.3
32.3
25.4 24.1
46.4 19.6
24.1 6.1 0.8
0.0 0.0 2.3
25.6
15.1
14.3
16.1
7.6
5.3
9.8
constant. The regression model included factors found to be significant in the univariate analyses as well as the matching variables used to define the strata as above. Adjusted odds ratios and confidence intervals were obtained from the logistic regression model. Since only a subgroup of subjects had ascorbic acid values available, this variable was added in a separate regression run. Pearson correlation analysis was used to examine whether the amount of smoking (cigarettes per day) was linearly related to the level of ascorbic acid. One value of ascorbic acid was clearly aberrant. Its value was equal to 9 mg/dl and the next largest value was equal to 1.8 mg/dl. Hence, this value was excluded in this analysis. The non-parametric Spearman correlation coefficient was also computed to insure that our findings did not depend on an as376 sumption of normal distributions.
The frequencies of positive study variables for the PPROM and control groups are listed in Table 1. Mantel-Haenszel tests of odds ratios were performed on study patients and matched controls. The factors achieving significance at the p < 0.05 level on univariate analysis are displayed in Table 2. A history of prior cerclage was noted to be a significant risk factor with a p value of less than 0.05. However, the prevalence was low so that this factor could not be included in the regression analysis described later. The odds ratio and confidence intervals were calculated using a 2-tailed Fisher's exact test. The odds ratio was found to be 7.33 with a 95% confidence interval of 1.11-167.68. Figure 1 demonstrates the dose response relationship observed with smoking. The relative proportion of cases of PPROM in the smoking level categories of 0, 1-10, 11-20, and 20 cigarettes per day with 40%, 58%, 78%, and 75%, respectively. A linear trend in the increase of these proportions was statistically significant (p < 0.001). Regression analysis revealed that only smoking greater than 10 cigarettes per day and a history of PPROM in previous pregnancies were significant independent predictors of PPROM as shown in Table 3. Twelve patients out of 266 were excluded from the multivariable analysis because of missing data. In the subgroup of patients (N = 34) and their matched controls on whom ascorbic acid levels were measured, an association between PPROM and a decreased ascorbic acid level (
October 1990
RISK FACTORS FOR PRETERM PREMATURE RUPTURE OF THE FETAL MEMBRANES Carolyn B. Hadley, M.D., Denise M. Main, M.D., and Steven G. Gabbe, M.D.
Preterm premature rupture of the membranes (PPROM) is a significant cause of prematurity, accounting for approximately one third of preterm births in the United States. PPROM occurs in approximately 0.7—2% of all pregnancies nationally, and has a reported recurrence rate of 21%. The elucidation of potential risk factors for PPROM could contribute to a better understanding of its etiology. To study the contributions of 20 potential risk factors, we undertook a case-control study in our clinic population, which has a 5-6% incidence of PPROM. One hundred and thirty-three patients experiencing PPROM were matched for race, age, parity and gestational age with undelivered patients. Studies performed included ultrasonographic examinations, blood levels of ascorbic acid and zinc, microbiologic assays, patient questionnaires, and chart reviews. After stratification of both groups into subgroups based on matching criteria, summary tests of significance and Mantel-Haenszel tests of odds ratios were performed. On univariate analysis the following factors achieved significance at the p < 0.05 level with 95% confidence intervals: 1) previous history of PPROM 2) smoking (dose related) 3) fundal location of the placenta in the present pregnancy. 4) a prior history of cerclage. After regression analysis, we concluded that smoking and history of previous PPROM were found to be risk factors for PPROM in our inner city black population.
The term "premature rupture of the membranes" (PROM) is generally accepted as referring to the leakage of amniotic fluid through the cervical os more than one hour prior to the onset of labor. Preterm PROM (PPROM) includes all cases of PROM occurring before 37 completed weeks gestational age. PPROM occurs in 0.7-2.0% l of all pregnancies, but accounts for approximately 30% of all preterm births2 and is the single most common diagnosis leading to admission to newborn intensive care nurseries. In a population-based study in North Carolina, Meis and his co-workers noted that PPROM is a much more frequent cause of preterm births in indigent patients than in private patients.3 In our clinic population of inner city black women, PPROM accounts for 5-6% of all deliveries, a figure 2-3 times the national average. Given the high incidence of PPROM among our population, we recognized the importance of identifying possible risk factors prospectively.
Many theories have been proposed to explain the etiology of PPROM. Most have focused on the disruption of the mechanical integrity of the membranes. 45 Damage to the membranes may be caused by substances produced by genital tract bacteria themselves or may result from the inflammatory response of the host.67 Decreased maternal levels of ascorbic acid, copper and zinc have been associated with PPROM.89 Smoking is known to impair protein metabolism and to reduce the levels of amino acids, Vitamin B12 and ascorbic acid.10 Meyer and Tonascia reported a three-fold increase of PPROM between 20-34 weeks' gestational age for smokers as opposed to non-smokers.l J Coitus has also been implicated as contributing to the incidence of PPROM12 in one study but not in others 1314 Toth et al implicated preexisting infection of the uterine cavity as a predisposing factor for premature rupture of the membranes and preterm delivery.15 The recurrence risk for PPROM has been noted to be as high as 21%. 16
From the Departments of Obstetrics and Gynecology, Hospital of the University of Pennsylvania, Medical College of Pennsylvania, University of California at San Francisco, and Ohio State University College of Medicine Reprint requests: Dr. Hadley, Dept. of Obstetrics and Gynecology, Medical College of Pennsylvania, 3300 Henry Avenue, Philadelphia, PA 19129 374
Copyright © 1990 by Thieme Medical Publishers, Inc., 381 Park Avenue South, New York, NY 10016. All rights reserved.
Downloaded by: Universite Laval. Copyrighted material.
ABSTRACT
Finally, a number of gynecologic procedures and pregnancy-related conditions have been thought to contribute to the occurrence of PPROM. Bleeding during the early part of the current pregnancy was found by Joffee to predispose patients to PPROM.17 Pregnancies with hydramnios tend to be more susceptible to PPROM, presumably because of mechanical stress on the cervix and membranes. Fundal placental location, which places the weakest point in the membranes over the cervical os, is felt to predispose patients to PPROM.18 Marginal cord insertions have been noted more frequently in patients with PPROM.19 To better elucidate the risk factors associated with PPROM and because of the absence of a controlled study in the literature which addresses most of the potential risk factors for this important perinatal complication, we undertook this case-control study.
MATERIALS AND METHODS
The design of this study, i.e. case control with non-random ascertainment of cases and controls, enabled the authors to work with a smaller study sample size than would have been required for a prospective study. Cases and controls were obtained sequentially. According to Schlesselman, this sample size with matched controls is adequate to test for a relative risk or 2.0 with Type I and II errors of 0.05 and 0.20, respectively.20 For the purposes of this study PPROM refers to rupture of the membranes occurring more than 1 hour prior to the onset of labor at a gestational age of less than 37 weeks. Gestational age was determined by ultrasonographic and clinical criteria. Rupture of membranes was diagnosed by the presence of a vaginal pool, a positive nitrazine test, microscopic ferning and by documenting decreased amniotic fluid on ultrasonographic exam. For one year, each patient on the Clinic Service at the Hospital of the University of Pennsylvania, Philadelphia, who was diagnosed as having PPROM was matched to an undelivered control patient who was being followed in the routine Obstetrical Clinic at the same institution. Undelivered controls were selected because it was felt that these patients were typical of our general obstetrical population. Patients considered to be at high risk by virtue of their medical and obstetrical histories were excluded from the control population. The controls were selected by the senior author by matching each new patient with PPROM to the next available patient in the routine obstetrical clinic whose matching parameters fit the necessary criteria. All study patients were black and had singleton gestations. The study subjects were matched to controls for age (greater or less than 20 yrs) parity (parous vs. nulliparous) and gestational age (within 1 —2 weeks). It was felt that such careful matching would limit confounding factors. The patients were matched for
age because of the potential difference in nutritional and social habits between teenagers and adults. The mean age for patients under 20 years was 17.8 years for the PPROM group and 17.9 years for controls. The mean age for patients over 20 years was 26.8 years for the PPROM group and 26.3 years for controls. Study subjects were matched for parity to take into account the potential contributions of the patient's prior obstetrical history. Patients were matched for gestational age because of the potentially greater influence of intrauterine infection at earlier gestational ages. All patients were under 37 weeks' gestational age. Every effort was made to match gestational ages to within one week with the outside limit being 2 weeks. There were 133 PPROM and 133 control patients recruited into the study which was approved by the Institutional Review Board. Written consent was obtained from each participating patient at the time of the interview. The patient questionnaire was administered as a structured interview by the senior author (CBH) to all participants. After delivery the answers in the questionnaire and medical history were verified by a thorough check of the entire medical record. Items covered in the questionnaire and medical record review are listed in Table 1. Each study patient underwent an ultrasonographic examination to determine placental location. In addition, cervical cultures for Neisseria gonorrhoeae, and a vaginal culture for Group B streptococcus were obtained in all patients. Serum ascorbic acid levels and plasma zinc levels were obtained in smaller groups of patients and their matched controls. The number of patients tested for zinc and ascorbic acid levels was limited by the cost of these assays; however, all matching criteria were fulfilled. Both specialized assays were performed in the laboratories of Smith-Kline Bioscience, Inc. (King of Prussia, PA). Cut-off values were provided by the laboratories and were as follows: ascorbic acid, 0.2 mg/dl and plasma zinc, 60 mcg/dl.
Statistical Methods
The univariate analyses of the effect of each potential risk factor took into account the matching scheme. Specifically, a cross classification of the matching variables produced strata in which the cases and controls were homogeneous with respect to the matching variables. The risk of PPROM was estimated within each strata; summary tests of significance and odds ratios, with their 95% confidence intervals, were obtained using the Mantel-Haenszel method. In addition to determining whether the presence of smoking was a risk factor, the authors tested for a linear dose response relationship using a chi-square test for linear trend. Multiple logistic regression was used to determine which factors contributed independently at least to some extent to elevating the risk of PPROM after controlling for other factors which were held 375
Downloaded by: Universite Laval. Copyrighted material.
PRETERM PREMATURE RUPTURE/Hadley, Main, Gabbe
AMERICAN JOURNAL OF PERINATOLOGY/VOLUME 7, NUMBER 4
October 1990
PPROM (%) N = 133 Historical factors Dilatation and curettage First trimester spontaneous abortions First trimester therapeutic abortions Second trimester spontaneous abortions Second trimester therapeutic abortions Previous preterm delivery without PPROM Previous PPROM Cere I age Cone biopsy Pelvic inflammatory disease requiring hospitalization Uterine malformations (by hysterosalpingogram) DES exposure (by history and exam) Current pregnancy Smoking (number of cigarettes per day) Coitus within 1 week Bleeding after twelve weeks No prenatal care Cerclage Hydramnios (pocket > 8 cm by ultrasound) Fundal placental location Microbial infection/colonization Group B streptococcus Neisseria gonorrhoeae
Control (%) N = 133
12.8 18.8
16.5
37.6
36.8
8.3
3.0
6.8
3.8
4.5
5.3
33.1
12.8
5.3 1.5 1.5
0.8 0.0 5.2
2.3
2.3
0.8
0.0
55.3
32.3
25.4 24.1
46.4 19.6
24.1 6.1 0.8
0.0 0.0 2.3
25.6
15.1
14.3
16.1
7.6
5.3
9.8
constant. The regression model included factors found to be significant in the univariate analyses as well as the matching variables used to define the strata as above. Adjusted odds ratios and confidence intervals were obtained from the logistic regression model. Since only a subgroup of subjects had ascorbic acid values available, this variable was added in a separate regression run. Pearson correlation analysis was used to examine whether the amount of smoking (cigarettes per day) was linearly related to the level of ascorbic acid. One value of ascorbic acid was clearly aberrant. Its value was equal to 9 mg/dl and the next largest value was equal to 1.8 mg/dl. Hence, this value was excluded in this analysis. The non-parametric Spearman correlation coefficient was also computed to insure that our findings did not depend on an as376 sumption of normal distributions.
The frequencies of positive study variables for the PPROM and control groups are listed in Table 1. Mantel-Haenszel tests of odds ratios were performed on study patients and matched controls. The factors achieving significance at the p < 0.05 level on univariate analysis are displayed in Table 2. A history of prior cerclage was noted to be a significant risk factor with a p value of less than 0.05. However, the prevalence was low so that this factor could not be included in the regression analysis described later. The odds ratio and confidence intervals were calculated using a 2-tailed Fisher's exact test. The odds ratio was found to be 7.33 with a 95% confidence interval of 1.11-167.68. Figure 1 demonstrates the dose response relationship observed with smoking. The relative proportion of cases of PPROM in the smoking level categories of 0, 1-10, 11-20, and 20 cigarettes per day with 40%, 58%, 78%, and 75%, respectively. A linear trend in the increase of these proportions was statistically significant (p < 0.001). Regression analysis revealed that only smoking greater than 10 cigarettes per day and a history of PPROM in previous pregnancies were significant independent predictors of PPROM as shown in Table 3. Twelve patients out of 266 were excluded from the multivariable analysis because of missing data. In the subgroup of patients (N = 34) and their matched controls on whom ascorbic acid levels were measured, an association between PPROM and a decreased ascorbic acid level (
