marijuana use during pregnancy and decreased length of gestation
TRANSCRIPT
Marijuana use during pregnancy and decreased
length of gestation
P. A. Fried, B. Watkinson, and A. Willan
Ottawa and Kingston, Ontario, Canada
In a prospective study the relationship of marijuana use during pregnancy to infant birth weight and length of gestation was examined in 583 women who delivered single live infants. Eighty-four women used
marijuana to varying degrees during pregnancy. Marijuana, alcohol, smoking, and nutritional habits were
estimated by repeated interviews. Compared to nonuse, an average use of marijuana six or more times per week during pregnancy was associated with a statistically significant reduction of 0.8 weeks in the length of gestation after consideration of the effects of nicotine, alcohol, parity, mother's prepregnancy
weight, and the sex of the infant. With similar adjustments no reduction in birth weight was noted. Among the heavy marijuana users the effect on gestation length was dose dependent. (AM. J. OBSTET. GVNECOL.
150:23, 1984.)
In contrast to the considerable body of scientific literature describing the effects of nicotine I and alcohoF when used during pregnancy, there is an almost total absence of data on the consequences of marijuana upon human fetal development. In the early 1970s it was estimated that among females in this age bracket 5% were currently using marijuana whereas by the end of the decade the precentage had doubled.3 Greenland et a1.4
, Fried," and Fried et al.6 reported that, at a conservative estimate, 13% of pregnant women use marijuana to some degree.
Recently the first descriptions of the behavior of infants born to women who had used marijuana during their pregnancy have been reported.5. 7 A significant increase in symptoms associated with nervous system abnormalities was observed in 2- to 3-day-old offspring of women who used marijuana more than five times a week. The purpose of the present paper is to report an association between regular marijuana use and a shortened gestation period.
Method and materials
Subjects. The subjects in this report were the infants in singleton births of 583 mothers-to-be who volunteered to participate after becoming informed of the study either by their obstetrician or by notices in the public media. The investigation was described as a study of prenatal life-style habits and their effects upon the infants. The women were interviewed once during
From the Department of Psychology, Carleton University, and Department of Community Health and Epidemiology, Queen's University.
Supported by a National Health and Welfare grant (P. A. F.). Received for publication May 23, 1983; revised November 21, 1983;
accepted March 21, 1984. Reprint requests: Dr. P. A. Fried, Department of Psychology, Carle
ton University, Ottawa, Ontario, Canada K1 S 5B6.
each of the trimesters remaining in their pregnancy. Information obtained during each interview included the volume and variability of alcohol consumption, cigarette smoking habits, marijuana intake, and the use of other prescribed and non prescribed drugs during the year before pregnancy and each trimester of pregnancy. In addition to this determination of drug usage, a general health history, details of previous pregnancies, a 24-hour dietary recall, and sociodemographic data were obtained.
Details of the interview have been described elsewhere.6 In that earlier report demographic information, dietary habits, and drug usage patterns of 217 women were described. The infants of these women are included in the sample of 583 described in the present paper.
The length of gestation was determined by the date of the last menstrual period. In cases where this information was not known precisely, anthropometric measurements and ultrasound data were used.
Due to the broad range and bimodal distribution of marijuana use among the women in the present sample, the drug data were not treated as a continuous variable but rather considered in terms of categories. As in earlier reports,5~7 the women were divided into
four categories with respect to marijuana use during pregnancy: nonusers; irregular users-those who smoked an average of no more than one marijuana
cigarette per week or were exposed to the exhaled smoke of others; moderate users-those whose use averaged between two and five marijuana cigarettes per week; heavy users-those whose average use was greater than five marijuana cigarettes per week.
For each outcome variable a linear model was developed with marijuana (in four categories) as the main factor. Considered as covariates were alcohol (ounces of absolute alcohol per day), nicotine use (categorized
23
24 Fried, Watkinson, and Willan September I, 1984 Am. J. Obstet. Gynecol.
Table I. Demographic data and nicotine and alcohol consumption among the four categories of marijuana users
Nonusers Irregular users Moderate users Heavy users Characteristics (n = 499) (n = 48) (n = 18) (n = 18)
Age at child's birth (mean, years) 29.3 26.0 26.4 25.9 Height (mean, cm) 161.5 159.6 160.7 161.7 Weight before pregnancy (mean, kg) 58.4 58.4 55.6 57.2 Parity (mean) 0.33 0.50 0.70 0.68 Education (graduated high school, %) 97 85 78 63 Family income (mean, dollars) 31,406 31,512 19.313 19,116 Smokers (>16 mg/day, %) 5 13 22 26.3 Alcohol (>0.85 ounce absolute 3 2 II 10.5
alcohol/day, %)
Table II. Unadjusted and adjusted gestation length and infant birth weight among the four categories of marijuana users
Irregular users Moderate users Heavy users Nonusers (n = 499) (n = 48) (n = 18) (n = 18)
I I I I Level of
Mean SD Mean SD Mean SD Mean SD significance
Gestation (weeks) Raw 39.7 1.67 40.2 1.19 40.2 1.83 38.8 1.77 0.008 Adjusted 39.6 40.2 40.2 38.8 0.008
Birth weight (gm) Raw 3490 523 3517 428 3512 425 3396 321 0.851 Adjusted 3482 3549 3599 3430 0.598
The raw means were examined with a one-way analysis of variance and the adjusted means with an analysis of covariance. Gestation length was adjusted for mother's prepregnancy weight. Birth weight was adjusted for maternal age, weight before pregnancy, sex of child, and nicotine use.
as nonsmokers, smokers of 1 to 15 mg of nicotine per day, and smokers of 16 or more milligrams per day), the mother's prepregnancy weight, parity, family income, mother's education, maternal age, and the sex of the infant. Maternal weight gain was purposely omitted from consideration because it is intermediate in the causal chain between the study factor (marijuana use)
and the study outcomes (gestation length and birth weight), and such intermediate variables should not be
controlled for. 8 For the same reason, gestation length was not considered as a covariate in the examination of the effect of marijuana use on birth weight.
For each outcome a final model was fitted by initially
including all factors and removing the least significant
factor one at a time with the provisions that the primary factor of marijuana use and any factor with a significance level of <0.15 were not removed. The final
model for gestation length included marijuana use and mother's prepregnancy weight. The final model for
birth weight included marijuana use, maternal age, mother's prepregnancy weight, the sex of the child, and nicotine use.
Results
Irregular use of marijuana in the first trimester was
reported by 7.7% of the women, moderate use by 1. 7%, and heavy use by 4.0%. In the third trimester the per-
centages of irregular, moderate, and heavy users were 6.3,0.7, and 3.0, respectively. Because the correlation of marijuana use across trimesters was high (r = 0.65 to 0.77, p < .001 in all three comparisons), the treatment of the data and the discussion of the findings are based on marijuana usage averaged throughout the course of pregnancy. It is recognized that since there is a large number of marijuana nonusers compared to the other three categories, the probability values associated with these correlation coefficients may be somewhat imprecise. Nonetheless, correlation coefficients are sensitive to associations between two variables regardless of the underlying distributions, and in this case the extreme
probability values indicate a high degree of association. The alternative approach of contingency table analyses
based on marijuana use in categories would suffer from small expected cell frequencies and inflated test
statistics. Of the 84 women who had used marijuana during
pregnancy, four reported using one other illicit drug (LSD or cocaine) and two reported amphetamine use. None reported having used these drugs beyond the second trimester. Samples of urine obtained during
the third trimester interview from all women who were moderate or heavy users of marijuana were screened for opiates and amphetamines,9 and none
were positive.
Volume 150 Number I
Since the use of marijuana, alcohol, and nicotine was
highly correlated (marijuana and nicotine: r = 0.329, P = 0.0001; marijuana and alcohol: r = 0.168, P = 0.0001; nicotine and alcohol: r = 0.157, P = 0.0001; Spearman's correlations), a multifactor analysis approach was adopted to examine the effect of the use of marijuana on the outcomes of gestation length and birth weight. The normality of distributions of the out
comes of gestation length and birth weight permitted the use of analysis of covariance. The variance of out
comes was homogeneous among marijuana groups (see Table II), and therefore the unequal group SIzes should not lead to biased results.
Demographic characteristics and alcohol and niCO
tine use in the four marijuana groups are shown in Table I. The 24-hour dietary recall data revealed no significant differences among any of the four groups. Although not statistically significant, the mean birth weight of infants born to the heavy marijuana users was 94 gm less than the mean birth weight of the infants
born to the nonusers. When birth weight was adjusted for the other factors in the model, this difference was slightly reduced to 78 gm (Table II).
The length of gestation was associated with mari
juana use even after covariate adjustment (p = 0.008). The mean length of gestation among the infants born to heavy users was 0.9 weeks less than for nonusers. After adjustment for the mother's prepregnancy
weight, the difference was only slightly reduced to 0.8 weeks (Table II). In the heavy user group, none of the pregnancies of <40 weeks was induced.
The analysis of covariance also indicated that (1) gestation length was positively related to the mother's prepregnancy weight (p = 0.001), (2) birth weight was positively related to maternal age (p = 0.055) and the mother's prepregnancy weight (p = 0.0001) and negatively related to nicotine use (p = 0.008), and (3) the male babies were heavier (p = 0.008).
The relationship between gestation length and marijuana use was investigated further after controlling for the mother's prepregnancy weight. With use of the Scheffe test for multiple comparisons1o at the 5% level, the mean residual gestation length of the heavy users was found to be significantly shorter than that of the
other two marijuana groups. A single comparison of
the heavy marijuana users and the nonusers, controlling for the mother's prepregnancy weight, showed
that the mean length of gestation of the heavy users
was significantly shorter than the mean of the nonusers (p = 0.039). Among the women categorized as heavy marijuana users a dose-dependent relationship (r = -0.401, P = 0.050, one-tail, Spearman's correlation)
between a shortened length of gestation and the amount of marijuana smoked was observed (Fig. 1). Spearman's rank correlations were used to eliminate
Marijuana use and decreased length of gestation 25
41
49
39 Gestation
Length (weeks) 38
37
36
35
34
T 1 I
* I I
.I.
6-12 (N=5)
T I I I I I ¥
.1.
13-19 (N-6)
T I
.1
29-59 (N-4 )
T I I I I
t I I
.J.
>59 (N-3 )
Average "joints"/week during pregnancy
Fig. 1. The dose-response relationship between marijuana used during pregnancy and length of gestation (r = -0.401, P = 0.050, one-tailed test). The range of gestation length is indicated for each grouping of heavy marijuana users.
the effect of one or two extreme values such as may occur in small samples.
The number of women within each category of
marijuana usage that gave birth to infants with a gesta
tion length of 36 weeks or less did not differ. The rate among the nonusers was 14 of 499 (3%), none were included in the 48 light users, one-at 36 weeks-was noted among the 18 moderate users (6%), and two
one at 34 weeks and the other at 36 weeks-were included among the 18 heavy users (11 %).
The reduced gestation length and absence of a small-for-dates effect in the marijuana users were in
marked contrast to those outcomes among the women who smoked cigarettes. As in previous reports (e.g., Able 1
) nicotine use in the present study was associated with lower birth weight but not with a shortened gestation period. A reduction of 232 gm was associated with a daily use of at least 16 mg of nicotine during pregnancy when adjusted for sex, mother's age, and weight before pregnancy. Alcohol usage at the social levels consumed by women in the present paper (only 14 drank a daily average of more than 0.85 ounce of absolute alcohol was not related to either reduced birth weight or an altered period of gestation.
Comment
Recently Hingson et al.,'l in a study conducted at the
Boston City Hospital, reported that reduced birth
weight but not a shortened length of gestation was as
sociated with maternal marijuana use during pregnancy. The 1690 women in that study were a young, low-income population of whom 20% reported eating less than three meals a day and a similar percentage incurred one or more serious illnesses during pregnancy. These and accompanying life-style differences between the Boston subjects and the women in our
26 Fried, Watkinson, and Willan
study (Table I) may be of fundamental importance in accounting for the differences obtained in the two studies. Animal work has shown that both inadequate and enriched nutritional factors can differentially interact with the consequences of fetal exposure to cannabisY The importance of the interaction of drug use during pregnancy and the general health/environment has been emphasized in alcohol,13 smoking,l and marijuana research. 14
Although there was a significantly lower income and level of formal education among the moderate and heavy marijuana users compared to the remainder of the sample in the present study, several factors warrant comment. There were no differences between the moderate and heavy users in terms of these dependent variables, but the reduced gestation length was observed only among the heavy marijuana users. Further, although income and education were reduced among the women categorized as moderate or heavy users, the average level of either of these variables was considerably higher than that found among inner city populations utilized in previous work. 11 Finally, in the present report both education and income were considered as covariates in the linear model that examined marijuana's role in gestation length .
Most of the women in the study of Hingson et al." were interviewed only once-immediately after giving birth. However, 328 women of a subsample were also interviewed during their pregnancy when they registered at the Boston City Hospital'S prenatal clinic. Unfortunately, information was not given as to how these women may have differed from the rest of the sample or when during their pregnancy the registration took place. An intriguing finding centered around the women given both the prenatal and postnatal interviews. In this subsample the marijuana reports were not consistent between the two interviews. A regression analysis was carried out to examine the effect of marijuana usage, as determined by the prenatal interview, upon birth weight and gestation length. In this case maternal marijuana habits were not related to infant size but were associated with a shorter length of gestation (p = 0.07). Thus the prenatal interview in the Boston study produced results that are consistent with those obtained in the present report (which also used prenatal interviews) whereas the postnatal retrospective interviews in the Boston study did not.
Recent animal work 15. 16 has indicated that tetrahydrocannabinol, the principal psychoactive constituent of marijuana, can alter a wide range of pituitaryovarian and adrenal hormones. There is also indirect evidence that tetrahydrocannabinol can affect steroid production by the placenta. 15. 16 Although the underlying biologic nature of a shortened gestation length associated with heavy cannabis use can only be speculated
September I , 1984 Am. J. Obstet. Gynecol.
upon at this time, a likely candidate for the mechanism of action is marijuana's influence upon the reproductive hormonal system.
It is of interest to note that over the centuries in many parts of the world marijuana has been reported to hasten childbirth, with observers noting that the drug increases the frequency and intensity of contractions. 17 The recent report of Greenland et a1.4 that precipitate labor was significantly more frequent among women who reported using marijuana "at least once a month" during pregnancy is consistent with the folk medicine observation and may be related to the findings of the present work.
Finally, although the finding of a significant doseresponse relationship between the heavy marijuana users and a shortened gestation period is suggestive, the results must be interpreted cautiously. An attempt was made in the present work to statistically control for nonmarijuana parameters that may have influenced the dependent variables of interest but there remains the possibility of other relevant confounding factors contributing to the observed association. Additional studies will be needed in order to isolate further the role of marijuana and gestation length, but as pointed out elsewhere,ll-14 even if the drug is identified as contributing to an aspect of pregnancy risk, it must be considered within a multifactor context. The drug is just one aspect of a life-style that, in total, may be contributing to a significant increase in the risk to the fetus.
The approximate I-week reduction in gestation length observed among the heavy marijuana users is of questionable clinical significance in and of itself. However, the figure of I week is an average, and the reduction is more marked as the quantity consumed increases. The finding that marijuana usage can contribute to a statistically shortened gestation length may take on clinical significance in certain individuals who consume large amounts of the drug and/or individuals whose life-style habits include the use of other drugs, such as alcohol,18 that may also reduce the length of gestation.
We thank M. Buckingham for her invaluable assistance and the staff of the Ottawa Civic, Ottawa General, Riverside, and Grace Hospitals fortheircooperation.
REFERENCES
1. Abel EL. Smoking during pregnancy: a review of effects on growth and development of offspring. Hum Bioi 1980;52:593.
2. Streissguth AP, Landesman-Dwyer S, Martin ]e, Smith OW. Teratogenic effects of alcohol in humans and laboratory animals. Science 1980;209:353.
3. National survey on drug abuse: 1979. Rockville, Maryland: National Institute on Drug Abuse, 1980.
4. Greenland S, Staisch K], Brown N, Gross Sj. The effects of marijuana use during pregnancy. I. A preliminary epidemiologic study. AM J OBSTET GYNECOL 1982;143: 408.
Volume 150 Number 1
5. Fried PA. Marihuana use by pregnant women and effects on offspring: an update. Neurobehav Toxicol Teratol 1982;4:451.
6. Fried PA, Watkinson B, Grant A, Knights RM. Changing patterns of soft drug use prior to and during pregnancy. Drug Alcohol Depend 1980;6:323.
7. Fried PA. Marihuana use by pregnant women: neurobehavioral effects in neonates. Drug Alcohol Depend 1980; 6:323.
8. MacManon B, Pugh TF. Epidemiology: principles and methods, Boston: Little, Brown & Co, 1970.
9. Marks V, Mould GP, O'Sullivan MJ, Teule JC. Monitoring of drug disposition by immunoassay. In: Bridges JW, Chasseaud LF, eds. Progress in drug metabolism, vol 5. New York: John Wiley, 1980:225-310.
10. Miller RG. Simultaneous statistical inference. New York : McGraw- Hill, 1981.
II. Hingson R, Alpert JJ , Day N, Dooling E, Kayne H , Morelock S, Oppenheimer E, Zuckerman B. Effects of maternal drinking and marijuana use on fetal growth and development. Pediatrics 1982;70:539.
Marijuana use and decreased length of gestation
12. Charlebois AT, Fried PA. Interactive effects of nutrition and cannabis upon rat perinatal development. Dev Psychobiol 1980; 13:591.
13. Rosett HL. The effects of alcohol on the fetus and offspring. In: Kalant OJ , ed: Alcohol and drug problems in women, vol. 5. New York: Plenum Press, 1980:595-652.
14. Fried PA, Buckingham M, Von Kulmiz P. Marihuana use during pregnancy and perinatal risk factors. AM J OBSTET GYNECOL 1983;146:992.
15. Abel EL. Marijuana and sex: a critical survey. Drug Alcohol Depend 1981 ;8: I.
16. Harclerode J. The effect of marijuana on reproduction and development. In : Peterson RC, ed: Marijuana research findings: 1980. Rockville, Maryland: National Institute on Drug Abuse, 1980: 137-166.
17. Abel EL. Marihuana: the first twelve thousand years. New York : Planum Press, 1980.
18. Berkowittz GS, Holford TR, Berkowitz RL. Effects of cigarette smoking, alcohol, coffee and tea consumption on preterm delivery. Early Hum Dev 1982;7:239.
Prostaglandin F2a , oxytocin, and uterine activation in
hypertonic saline-induced abortions
Anna-Riitta Fuchs, Anne B. Rasmussen, Jaana Rehnstrom, and Miklos Toth
New York, New York
Intra-amniotic injections of hypertonic saline at midgestation induce uterine activity, which evolves into a
laborlike pattern in less than 24 hours and is associated with progressive increase in uterine oxytocin
response . This uterine activation occurred in the absence of a measurable increase in plasma 13,14-dihydro-15-keto-prostaglandin F2a (PGFM). Only after 25 to 27 hours was a rise in plasma
concentration of PGFM detected, which then increased in a parallel manner with cervical dilatation. By
contrast, plasma oxytocin levels increased by almost 100% soon after the injection of hypertoniC saline, declining to initial levels by 24 to 28 hours. OxytOCin infusions given after the intra-amniotic injection at
rates resulting in about a fivefold increase in plasma oxytocin significantly accelerated cervical dilatation
and the rise in plasma PGFM. Oxytocin infused before induction of abortion resulted in only a small and transient rise in plasma PGFM. Hypertonic saline injections thus increase the prostaglandin F,astimulating action of oxytocin, which in turn may be responsible for the enhanced contractile response to the hormone. Myometrial activation after hypertonic saline injections is probably caused by an interaction
of oxytocin and prostaglandin F2a, and cervical dilatation depends on contractile activity and a critical
increase in prostaglandin production. (AM . J . OesTET. GVNECOL. 150:27, 1984.)
Prostaglandins E2 and F2a are believed to play an important role in the activation of human uterine con-
From the Department of Obstetrics and Gynecology, Cornell University M edical College.
This study was supported in part by grants from Sandoz Foundation (A. B. R.), The Finnish Academy of Sciences, and Finska Lakarsallskapet U. R.). The antiserum against PGFM was a gift from Dr. K. E . Kirton, The Upjohn Company, Kalamazoo, Michigan.
Received for publication August 16, 1983; revised February 15, 1984; accepted March 21,1984.
Reprint requests: Dr. Anna-Riitta Fuchs, Room 5-412, Department of Obstetrics and Gynecology, Cornell University Medical College, 1300 York Ave., New York, NY 10021.
tractions during labor, but it has not been possible to
establish with certainty whether they are essential for the initiation of labor. Although it is well established
that the concentration of prostaglandin F2a metabolites rises during labor, the rise does not become significant until relatively late in labor, when the cervix is >5 em dilated." 2 Prostaglandin E and F concentrations also
rise in the amniotic fluid in the course of labor, but a significant rise over late pregnancy values has not been detected in early labor.3 Usually the rise in amniotic fluid and plasma prostaglandin levels parallels the increase in cervical dilatation. The increased production
27