1373
eclampsia include vasoconstriction, plasma volume
depletion, and excessive salt retention.16 The consequence of these changes with respect to the circulation in the kidney may provide an alternative explanation for urate retention. The undiscovered factor or factors that cause pre-eclamptic vasospasm may exert an effect on renal tubular function similar to that of angiotensin ii which is known to depress urate clearance without lowering G.F.R. 17 Further work needs to be completed before these relations can be clarified. The value of measuring plasma-urate concentrations in hypertensive pregnancy is greatest between 24 and 32 weeks’ gestation. Low values indicate a good prognosis for the fetus. Rising or high values at this time define a small number of very high-risk cases which are better managed and treated in hospital. Early bed-rest, monitoring of fetal wellbeing in utero, and anticipation of maternal problems related to pre-eclampsia then ensure the best chances for bringing the pregnancy to a stage where planned delivery both prevents serious maternal complications and gives the best possible chance of fetal survival. The measurement is thus of great diagnostic value. There is however no evidence from these data that there is a particular plasma-urate concentration above which fetal death becomes immediately imminent. Therefore the prediction of impending fetal death and the exact timing of delivery must rely on other measures of placental and fetal function. Requests for reprints should be addressed to C. W. G. R. REFERENCES 1.
Butler,
N.
R., Bonham,
D. G.
(editors)
in Perinatal
Mortality; p. 87,
Edin
burgh, 1963. 2. Slemons, J. M., Bogert, L. J. J. biol. Chem. 1917, 32, 63. 3. Chesley, L. C., Williams, L. O. Am. J. Obstet. Gynec. 1945, 50, 367. 4. Seitchik, J. ibid. 1953, 65, 981. 5. Lancet, M., Fisher, I. L. J. Obstet. Gynœc. Br. Commonw. 1956, 63, 116 6. Stander, H. J., Cadden, J. F. Am. J. Obstet. Gynec. 1934, 28, 856. 7. Pollak, V. E., Nettles, J. B. Medicine, 1960, 39, 469. 8. Connon, A. F., Wadsworth, R. J. Aust. N. Z. J. Obstet. Gynœc. 1968, 8, 197 9. McFarlane, C. N. J. Obstet. Gynœc. Br. Commonw. 1970, 70, 63. 10. Pritchard, J. A., Stone, S. R. Am. J. Obstet. Gynec. 1967, 99, 754. 11. Suonio, S., Nummi, S., Jouppila, P., Reinila, M. Ann. chir. gynœc. fenn 1972, 61, 202. 12. Beilin, L. J., Redman, C. W. G., Bonnar, J. in Tenth Symposium on Advanced Medicine. (edited by J. G. Ledingham), p.1, London, 1974. 13. Redman, C. W. G., Beilin, L. J., Bonnar, J. J. clin. Path. (in the press). 14. Little, W. Obstet. Gynec., N.Y. 1960, 15, 109. 15. Handler, J. S. J. clin. Invest. 1960, 39, 1526. 16. Sims, E. A. Am. J. Obstet. Gynec. 1970, 107, 154. 17. Ferris, T. F., Gordon, P. Am. J. Med. 1968, 44, 359.
ASPIRIN AND CONGENITAL MALFORMATIONS DENNIS SLONE OLLI P. HEINONEN DAVID W. KAUFMAN
VICTOR SISKIND RICHARD R. MONSON SAMUEL SHAPIRO
Drug Epidemiology Unit, Boston University Medical Center, Cambridge, Massachusetts 02138, U.S.A.
Summary
cohort of 50 282 gravidas and their offspring in the U.S.A., malformation rates were similar in the children of 35 418 women not exposed to aspirin, 9736 with intermediate exposure, and 5128 women heavily exposed during the first four lunar months of pregnancy. After controlling a wide range of potential confounding factors using multivariate analysis, the observed and expected numbers for In
a
a variety of malformation categories were similar in all three comparison groups. The data suggest that aspirin is not teratogenic.
Introduction ASSOCIATIONS between aspirin taken during pregnancy and congenital malformations, 1low birthweight,34 and a high perinatal mortality-rate3 4 have been reported. The widespread use and availability of the drug gives rise to particular concern over these findings. Here we report on the possible teratogenicity of aspirin. The accompanying report5 evaluates aspirin use ;n relation to birth-weight and perinatal mortality.
Methods The data described here were obtained from a prospective cohort study, the Collaborative Perinatal Project, which has been described elsewhere.6-8 A cohort of 50 282 mother-child pairs, seen in 12 hospitals throughout the U.S.A., was selected for study. Before the birth of the child, information was collected on drugs taken during pregnancy, maternal illnesses, complications of pregnancy, and other factors. Data on drug use were recorded at each antenatal visit, and were confirmed, with few exceptions, by the attending physician or by review of the hospital or clinic record. Full details of dosage were not collected; the heaviest exposure category recorded was for 8 or more days in any lunar month. Aspirin was the most commonly used drug, being taken at some time during pregnancy by 32 164 women. In evaluating aspirin in relation to congenital malformations, attention was confined to the first four lunar months of pregnancy-i.e., the period covering early embryogenesis. Aspirin was taken by 14 864 women during this period. Because of the way in which the data were collected, "heavy exposure" was defined as aspirin taken for at least eight days during at least one of the first four lunar months. There were 5128 heavily exposed mother-child pairs. 3248 children (6.5%), including survivors and those who died, had one or more congenital malformations identified before the first birthday or at death before the fourth birthday. Of these, 1128 had one or more of six malformations (inguinal hernia, 683 children; clubfoot, 192; cleft gum, 110; pectus excavatum, 92; urethral obstruction, 68; abnormal hands and fingers, 10) which showed considerable variability in the rates among the 12 hospitals (non-uniform malformations). Rates among the hospitals were reasonably uniform for all other malformations, which affected 2277 children (uniform malformations) ; 157 children had both uniform and non-uniform malformations and were includedin the uniform group. Of the 2277 children with uniform malformations, 1393 had malformations of "major" severity (2.8%).9 All 2277 children and the subgroup of 1393 with major malformations were independently evaluated, as were ten further subgroups, eight of which were based on the anatomical location of the deformity (e.g., central nervous, cardiovascular, respiratory); the two remaining subgroups were syndromes and tumours. Many of the children had multiple lesions, and the subgroups were not mutually exclusive. Of the children with non-uniform malformations, only those with inguinal hernia and clubfoot are considered here. Data concerning the remaining four non-uniform outcomes will be reported elsewhere.8 In order to control possible confounding, risk-factors for each of the outcomes were identified. A detailed description of the risk-factors and of the methods used to identify them will be presented elsewhere.8 Briefly, comparisons were made between children with and without each of the outcomes in terms of certain factors. These factors included: administrative data, such as the number of antenatal visits; personal characteristics of the mothers, such as age; characteristics of the offspring, such as birth-weight; reproductive history, such as
1374 TABLE I—CONGENITAL MALFORMATIONS ACCORDING TO ASPIRIN EXPOSURE DURING EARLY PREGNANCY
*
(LUNAR MONTHS 1-4)
in children who were heavily exposed to aspirin were 1.08 and 1.11 respectively. The largest relative risk, ’1-21 for gastrointestinal malformations, wasnot accounted for by an excess of any specific deformity. With regard to any exposure to aspirin (whether heavy or not), the standardised relative risks of uniform and major malformations were 1.00 and 1.01 respectively, with approximate upper 95% confidence limits of 1.06 and 1.09.
Reference category
Discussion
parity; illnesses, such as diabetes; complications of pregnancy, as pre-eclampsia; environmental factors, such as cigarette smoking; and genetic factors, such as prior malformed siblings. Factors found in the initial comparisons to be associated such
with the outcome under consideration were selected for further evaluation by means of linear discriminant function analysis. Certain factors were eliminated at that point by their covariates (e.g., low birth-weight frequently eliminated low placental weight). Those factors not eliminated were then evaluated by multiple logistic risk function analysis.8 Separate models, having different sets of risk factors, were developed for each outcome. Use of the multiple logistic risk function models made it possible to assign to each mother-child pair a score which represented that pair’s risk of any given outcome. For any drug exposure group, summing the scores yielded an estimate of the expected number of malformed children, in the absence of drug exposure. The ratio of the observed number of malformed children exposed to a drug (in this case, aspirin) to the expected number gave a standardised relative risk that simultaneously took into account potential confounding from all identified risk factors.
Results The overall malformation rates were similar in 5128 heavily exposed children, 9736 other exposed children, and in 35 418 children who were not exposed to aspirin during the first four lunar months of pregnancy (table
I). When the children were divided according to the various outcomes, unstandardised rates in both aspirin exposure groups were similar to those observed in the unexposed group (table II). All standardised relative risks approximated unity. The upper approximate 95% confidence limits for uniform and major malformations
The results of this study give no evidence that aspirin ingestion during pregnancy is associated with congenital
malformations. Furthermore, the upper 95% confidence limits of the relative risk estimates for uniform malformations and for major malformations make it unlikely that substantial teratogenic effects would have escaped detection. If aspirin is in fact a cause of malformations, several factors could have obscured evidence to that effect in this study. First, chance may explain failure to detect relationships with some of the less common outcomes. Second, even though multiple logistic risk function analysis was used to simultaneously control a wide range of potential confounding factors, the possibility of negative confounding by undetected factors could not be rulled out. Third, a systemic bias in the data collection ’could have obscured an association. Certainly, observer bias was unlikely in this study because the information on drug exposure was collected before delivery. Fourth, some degree of underestimation of aspirin use was undoubtedly present, since the median time of entry into the study was 21.6 wk: some women may not have recalled taking aspirin during early pregnancy. However, there was less likelihood of underestimation among heavy users. In addition, misclassification of aspirin users as non-users would have had to be very common to completely obscure an actual association, because the non-exposed group was extremely large. The data presented here do not accord with two previous studies.’In both studies, the interviewers had prior knowledge of the hypothesis when determining whether or not the subjects had used salicylates.
TABLE II—CATEGORIES OF CONGENITAL MALFORMATIONS ACCORDING TO ASPIRIN EXPOSURE DURING EARLY PREGNANCY
*
95% confidence limits: 0.84-1.08
t 95% confidence limits: 0.80—1.11 t Excludes polydactyly in Black children and clubfoot
Excludes ingumal herma t Male fetuses exposed: 2650 lExcludes Down syndrome
(LUNAR MONTHS 1-4)
heavy; 4929 other;
17 963
not
exposed
1375 It remains possible that grossly excessive exposure to aspirin may be teratogenic. However, based on limited data, such an effect was not evident in the offspring of Australian women whotook unusually large amounts of salicylates.3 Based on a larger body of data, more conventional doses of aspirin as used by pregnant American women do not appear to cause malformations in their
offspring. The following institutions participated in the Collaborative Perinatal Project of the National Institute of Neurological Diseases and Stroke: Boston Lying-In Hospital; Brown University; Charity Hospital, New Orleans; Children’s Hospital of Buffalo; Children’s Hospital of Philadelphia; Children’s Hospital Medical Center, Boston; Columbia-Presbyterian Medical Center; Johns Hopkins Hospital; Medical College of Virginia; New York Medical College; Pennsylvania Hospital ; University of Minnesota Hospitals; University of Oregon Medical School; and University of Tennessee. This work was supported by contract 223-75-3036 from the Food and Drug Administration, contract NOI-NS-2-2322 from the National Institute of Neurological Diseases and Stroke, and a grant from Hoffman-LaRoche, Inc., in Nutley, New Jersey.
Requests logy Unit,
for reprints should be addressed to S.S., Drug Epidemio10 Moulton Street, Cambridge, Massachusetts 02138,
U.S.A.
REFERENCES
1. Richards, I. D. G. Br. J. prev. soc. Med 1969, 23, 218. 2. Nelson, M. M., Forfar, J. O. Br. med. J. 1971, i, 523. 3. Turner, G., Collins, E. Lancet, 1975, ii, 338. 4. Collins, E., G. ibid. p. 335. 5. Shapiro, S., Siskind, V., Monson, R. R., Heinonen, O. P., Kaufman, D. W. Slone, D. Lancet, 1972, i, 1375. 6. Nieswander, K. R., Gordon, M. (editors) The Women and Their Pregnancies; Philadelphia, 1972. 7. Slone, D., Heinonen, O. P., Monson, R. R., Shapiro, S., Hartz, S. C., Rosenberg, L. Clin. Pharmac. Ther. 1973, 14, 648. 8. Heinonen, O. P., Slone, D., Shapiro, S. Birth Defects and Drugs in Pregnancy (in the press). 9. Shapiro, S. Slone, D., Hartz, S. C., Rosenberg, L., Siskind, V., Monson R. R., Mitchell, A., Heinonen, O. P., Idänpään-Heikkilä, Härö, S., Saxén, L. Lancet, 1976, i, 272.
PERINATAL MORTALITY AND BIRTH-WEIGHT IN RELATION TO ASPIRIN TAKEN DURING PREGNANCY SAMUEL SHAPIRO RICHARD R. MONSON DAVID W. KAUFMAN
VICTOR SISKIND OLLI P. HEINONEN DENNIS SLONE
Drug Epidemiology Unit, Boston University Medical Center, Cambridge, Massachusetts 02138, U.S.A. cohort of 41 337 gravidas and their offspring in the U.S.A. there was no evidence that aspirin taken in pregnancy is a cause of stillbirth, neonatal death, or reduced birth-weight. The women were divided into those who were not exposed to aspirin (14 956), those with intermediate exposure (24 866), and those who were heavily exposed (1515). Stillbirth-rates were similar for all three groups. Differences in neonatal death-rates and mean birth-weights were slight and none were statistically significant.
Summary
Trends
were
In
a
opposite
among
approximately equal
numbers of White and Black children. Introduction THE results of a recent study suggest that salicylates during pregnancy may reduce birth-weight and
taken
increase the risk of perinatal death. 12 We report vestigation into these issues.
an
in-
Methods The Collaborative Perinatal Project is described in the prethe purposes of the present study, the cohort of 50 282 mother-child pairs was modified as follows: (1) When a mother was enrolled in the study more than once, a random pregnancy was selected. This was done because perinatal deaths in prior siblings may increase the risk of subsequent perinatal death. (2) Pregnancies lasting less than 7 lunar months were excluded, since as explained below, the definition of heavy aspirin exposure used here was partly dependent upon the duration of pregnancy. The reduced cohort consisted of 41 337 mother-child pairs, among which there were 520 stillbirths (1.3%) and 437 neo-
ceding papery For
natal deaths (11%). 26 381 women used
aspirin at some time during pregnancy. defined, for pregnancies lasting at least "Heavy exposure" eight lunar months, as aspirin taken for at least eight days per lunar month in at least six lunar months. For pregnancies of seven lunar months duration, the drug had to be taken for at least eight days in each of at least five lunar months. There were 1515 heavily exposed mother-child pairs. Aspirin was less commonly used by non-smokers, nulliparz, was
White women, and normotensives. In addition, because of the way in which heavy exposure was defined in this study, such exposure was more common in pregnancies that lasted longer. Cigarette-smoking and parity were also related to perinatal mortality. Therefore perinatal mortality-rates were standardised by the direct method for these two factors. Birthweight was related to duration of pregnancy, cigarette-smoking, birth order, and pre-existing hypertension complicating pre-eclampsia. Mean birth-weights were adjusted by linear regression techniques for these factors. The standardised analyses were carried out separately for the White and Black ethnic groups. Because of small numbers, 3124 Puerto Rican mother-child pairs were included only when considering overall unadjusted perinatal mortality-rates (table I). In all further analyses they were excluded, leaving 38 213 mother-child pairs.
Results shows the stillbirth and neonatal death-rates to aspirin exposure during pregnancy. The rates were similar for each category of exposure. In table n, standardised stillbirth and neonatal deathrates are shown according to ethnic group. Among White children the stillbirth-rates were similar for all categories of exposure. Among Black children, the stillbirth-rates decreased with increasing antenatal exposure to aspirin, but this trend was not statistically significant. The neonatal death-rate was higher in White children exposed in utero to heavy aspirin-use than in non-exposed children. However, among heavily exposed Black children, the neonatal death-rate was lower. Neither of the differences was statistically significant. Other exposure to aspirin was not associated with neonatal death.
Table
i
according
TABLE I-STILLBIRTHS AND NEONATAL DEATHS IN RELATION TO ASPIRIN EXPOSURE DURING PREGNANCY
eclampsia include vasoconstriction, plasma volume
depletion, and excessive salt retention.16 The consequence of these changes with respect to the circulation in the kidney may provide an alternative explanation for urate retention. The undiscovered factor or factors that cause pre-eclamptic vasospasm may exert an effect on renal tubular function similar to that of angiotensin ii which is known to depress urate clearance without lowering G.F.R. 17 Further work needs to be completed before these relations can be clarified. The value of measuring plasma-urate concentrations in hypertensive pregnancy is greatest between 24 and 32 weeks’ gestation. Low values indicate a good prognosis for the fetus. Rising or high values at this time define a small number of very high-risk cases which are better managed and treated in hospital. Early bed-rest, monitoring of fetal wellbeing in utero, and anticipation of maternal problems related to pre-eclampsia then ensure the best chances for bringing the pregnancy to a stage where planned delivery both prevents serious maternal complications and gives the best possible chance of fetal survival. The measurement is thus of great diagnostic value. There is however no evidence from these data that there is a particular plasma-urate concentration above which fetal death becomes immediately imminent. Therefore the prediction of impending fetal death and the exact timing of delivery must rely on other measures of placental and fetal function. Requests for reprints should be addressed to C. W. G. R. REFERENCES 1.
Butler,
N.
R., Bonham,
D. G.
(editors)
in Perinatal
Mortality; p. 87,
Edin
burgh, 1963. 2. Slemons, J. M., Bogert, L. J. J. biol. Chem. 1917, 32, 63. 3. Chesley, L. C., Williams, L. O. Am. J. Obstet. Gynec. 1945, 50, 367. 4. Seitchik, J. ibid. 1953, 65, 981. 5. Lancet, M., Fisher, I. L. J. Obstet. Gynœc. Br. Commonw. 1956, 63, 116 6. Stander, H. J., Cadden, J. F. Am. J. Obstet. Gynec. 1934, 28, 856. 7. Pollak, V. E., Nettles, J. B. Medicine, 1960, 39, 469. 8. Connon, A. F., Wadsworth, R. J. Aust. N. Z. J. Obstet. Gynœc. 1968, 8, 197 9. McFarlane, C. N. J. Obstet. Gynœc. Br. Commonw. 1970, 70, 63. 10. Pritchard, J. A., Stone, S. R. Am. J. Obstet. Gynec. 1967, 99, 754. 11. Suonio, S., Nummi, S., Jouppila, P., Reinila, M. Ann. chir. gynœc. fenn 1972, 61, 202. 12. Beilin, L. J., Redman, C. W. G., Bonnar, J. in Tenth Symposium on Advanced Medicine. (edited by J. G. Ledingham), p.1, London, 1974. 13. Redman, C. W. G., Beilin, L. J., Bonnar, J. J. clin. Path. (in the press). 14. Little, W. Obstet. Gynec., N.Y. 1960, 15, 109. 15. Handler, J. S. J. clin. Invest. 1960, 39, 1526. 16. Sims, E. A. Am. J. Obstet. Gynec. 1970, 107, 154. 17. Ferris, T. F., Gordon, P. Am. J. Med. 1968, 44, 359.
ASPIRIN AND CONGENITAL MALFORMATIONS DENNIS SLONE OLLI P. HEINONEN DAVID W. KAUFMAN
VICTOR SISKIND RICHARD R. MONSON SAMUEL SHAPIRO
Drug Epidemiology Unit, Boston University Medical Center, Cambridge, Massachusetts 02138, U.S.A.
Summary
cohort of 50 282 gravidas and their offspring in the U.S.A., malformation rates were similar in the children of 35 418 women not exposed to aspirin, 9736 with intermediate exposure, and 5128 women heavily exposed during the first four lunar months of pregnancy. After controlling a wide range of potential confounding factors using multivariate analysis, the observed and expected numbers for In
a
a variety of malformation categories were similar in all three comparison groups. The data suggest that aspirin is not teratogenic.
Introduction ASSOCIATIONS between aspirin taken during pregnancy and congenital malformations, 1low birthweight,34 and a high perinatal mortality-rate3 4 have been reported. The widespread use and availability of the drug gives rise to particular concern over these findings. Here we report on the possible teratogenicity of aspirin. The accompanying report5 evaluates aspirin use ;n relation to birth-weight and perinatal mortality.
Methods The data described here were obtained from a prospective cohort study, the Collaborative Perinatal Project, which has been described elsewhere.6-8 A cohort of 50 282 mother-child pairs, seen in 12 hospitals throughout the U.S.A., was selected for study. Before the birth of the child, information was collected on drugs taken during pregnancy, maternal illnesses, complications of pregnancy, and other factors. Data on drug use were recorded at each antenatal visit, and were confirmed, with few exceptions, by the attending physician or by review of the hospital or clinic record. Full details of dosage were not collected; the heaviest exposure category recorded was for 8 or more days in any lunar month. Aspirin was the most commonly used drug, being taken at some time during pregnancy by 32 164 women. In evaluating aspirin in relation to congenital malformations, attention was confined to the first four lunar months of pregnancy-i.e., the period covering early embryogenesis. Aspirin was taken by 14 864 women during this period. Because of the way in which the data were collected, "heavy exposure" was defined as aspirin taken for at least eight days during at least one of the first four lunar months. There were 5128 heavily exposed mother-child pairs. 3248 children (6.5%), including survivors and those who died, had one or more congenital malformations identified before the first birthday or at death before the fourth birthday. Of these, 1128 had one or more of six malformations (inguinal hernia, 683 children; clubfoot, 192; cleft gum, 110; pectus excavatum, 92; urethral obstruction, 68; abnormal hands and fingers, 10) which showed considerable variability in the rates among the 12 hospitals (non-uniform malformations). Rates among the hospitals were reasonably uniform for all other malformations, which affected 2277 children (uniform malformations) ; 157 children had both uniform and non-uniform malformations and were includedin the uniform group. Of the 2277 children with uniform malformations, 1393 had malformations of "major" severity (2.8%).9 All 2277 children and the subgroup of 1393 with major malformations were independently evaluated, as were ten further subgroups, eight of which were based on the anatomical location of the deformity (e.g., central nervous, cardiovascular, respiratory); the two remaining subgroups were syndromes and tumours. Many of the children had multiple lesions, and the subgroups were not mutually exclusive. Of the children with non-uniform malformations, only those with inguinal hernia and clubfoot are considered here. Data concerning the remaining four non-uniform outcomes will be reported elsewhere.8 In order to control possible confounding, risk-factors for each of the outcomes were identified. A detailed description of the risk-factors and of the methods used to identify them will be presented elsewhere.8 Briefly, comparisons were made between children with and without each of the outcomes in terms of certain factors. These factors included: administrative data, such as the number of antenatal visits; personal characteristics of the mothers, such as age; characteristics of the offspring, such as birth-weight; reproductive history, such as
1374 TABLE I—CONGENITAL MALFORMATIONS ACCORDING TO ASPIRIN EXPOSURE DURING EARLY PREGNANCY
*
(LUNAR MONTHS 1-4)
in children who were heavily exposed to aspirin were 1.08 and 1.11 respectively. The largest relative risk, ’1-21 for gastrointestinal malformations, wasnot accounted for by an excess of any specific deformity. With regard to any exposure to aspirin (whether heavy or not), the standardised relative risks of uniform and major malformations were 1.00 and 1.01 respectively, with approximate upper 95% confidence limits of 1.06 and 1.09.
Reference category
Discussion
parity; illnesses, such as diabetes; complications of pregnancy, as pre-eclampsia; environmental factors, such as cigarette smoking; and genetic factors, such as prior malformed siblings. Factors found in the initial comparisons to be associated such
with the outcome under consideration were selected for further evaluation by means of linear discriminant function analysis. Certain factors were eliminated at that point by their covariates (e.g., low birth-weight frequently eliminated low placental weight). Those factors not eliminated were then evaluated by multiple logistic risk function analysis.8 Separate models, having different sets of risk factors, were developed for each outcome. Use of the multiple logistic risk function models made it possible to assign to each mother-child pair a score which represented that pair’s risk of any given outcome. For any drug exposure group, summing the scores yielded an estimate of the expected number of malformed children, in the absence of drug exposure. The ratio of the observed number of malformed children exposed to a drug (in this case, aspirin) to the expected number gave a standardised relative risk that simultaneously took into account potential confounding from all identified risk factors.
Results The overall malformation rates were similar in 5128 heavily exposed children, 9736 other exposed children, and in 35 418 children who were not exposed to aspirin during the first four lunar months of pregnancy (table
I). When the children were divided according to the various outcomes, unstandardised rates in both aspirin exposure groups were similar to those observed in the unexposed group (table II). All standardised relative risks approximated unity. The upper approximate 95% confidence limits for uniform and major malformations
The results of this study give no evidence that aspirin ingestion during pregnancy is associated with congenital
malformations. Furthermore, the upper 95% confidence limits of the relative risk estimates for uniform malformations and for major malformations make it unlikely that substantial teratogenic effects would have escaped detection. If aspirin is in fact a cause of malformations, several factors could have obscured evidence to that effect in this study. First, chance may explain failure to detect relationships with some of the less common outcomes. Second, even though multiple logistic risk function analysis was used to simultaneously control a wide range of potential confounding factors, the possibility of negative confounding by undetected factors could not be rulled out. Third, a systemic bias in the data collection ’could have obscured an association. Certainly, observer bias was unlikely in this study because the information on drug exposure was collected before delivery. Fourth, some degree of underestimation of aspirin use was undoubtedly present, since the median time of entry into the study was 21.6 wk: some women may not have recalled taking aspirin during early pregnancy. However, there was less likelihood of underestimation among heavy users. In addition, misclassification of aspirin users as non-users would have had to be very common to completely obscure an actual association, because the non-exposed group was extremely large. The data presented here do not accord with two previous studies.’In both studies, the interviewers had prior knowledge of the hypothesis when determining whether or not the subjects had used salicylates.
TABLE II—CATEGORIES OF CONGENITAL MALFORMATIONS ACCORDING TO ASPIRIN EXPOSURE DURING EARLY PREGNANCY
*
95% confidence limits: 0.84-1.08
t 95% confidence limits: 0.80—1.11 t Excludes polydactyly in Black children and clubfoot
Excludes ingumal herma t Male fetuses exposed: 2650 lExcludes Down syndrome
(LUNAR MONTHS 1-4)
heavy; 4929 other;
17 963
not
exposed
1375 It remains possible that grossly excessive exposure to aspirin may be teratogenic. However, based on limited data, such an effect was not evident in the offspring of Australian women whotook unusually large amounts of salicylates.3 Based on a larger body of data, more conventional doses of aspirin as used by pregnant American women do not appear to cause malformations in their
offspring. The following institutions participated in the Collaborative Perinatal Project of the National Institute of Neurological Diseases and Stroke: Boston Lying-In Hospital; Brown University; Charity Hospital, New Orleans; Children’s Hospital of Buffalo; Children’s Hospital of Philadelphia; Children’s Hospital Medical Center, Boston; Columbia-Presbyterian Medical Center; Johns Hopkins Hospital; Medical College of Virginia; New York Medical College; Pennsylvania Hospital ; University of Minnesota Hospitals; University of Oregon Medical School; and University of Tennessee. This work was supported by contract 223-75-3036 from the Food and Drug Administration, contract NOI-NS-2-2322 from the National Institute of Neurological Diseases and Stroke, and a grant from Hoffman-LaRoche, Inc., in Nutley, New Jersey.
Requests logy Unit,
for reprints should be addressed to S.S., Drug Epidemio10 Moulton Street, Cambridge, Massachusetts 02138,
U.S.A.
REFERENCES
1. Richards, I. D. G. Br. J. prev. soc. Med 1969, 23, 218. 2. Nelson, M. M., Forfar, J. O. Br. med. J. 1971, i, 523. 3. Turner, G., Collins, E. Lancet, 1975, ii, 338. 4. Collins, E., G. ibid. p. 335. 5. Shapiro, S., Siskind, V., Monson, R. R., Heinonen, O. P., Kaufman, D. W. Slone, D. Lancet, 1972, i, 1375. 6. Nieswander, K. R., Gordon, M. (editors) The Women and Their Pregnancies; Philadelphia, 1972. 7. Slone, D., Heinonen, O. P., Monson, R. R., Shapiro, S., Hartz, S. C., Rosenberg, L. Clin. Pharmac. Ther. 1973, 14, 648. 8. Heinonen, O. P., Slone, D., Shapiro, S. Birth Defects and Drugs in Pregnancy (in the press). 9. Shapiro, S. Slone, D., Hartz, S. C., Rosenberg, L., Siskind, V., Monson R. R., Mitchell, A., Heinonen, O. P., Idänpään-Heikkilä, Härö, S., Saxén, L. Lancet, 1976, i, 272.
PERINATAL MORTALITY AND BIRTH-WEIGHT IN RELATION TO ASPIRIN TAKEN DURING PREGNANCY SAMUEL SHAPIRO RICHARD R. MONSON DAVID W. KAUFMAN
VICTOR SISKIND OLLI P. HEINONEN DENNIS SLONE
Drug Epidemiology Unit, Boston University Medical Center, Cambridge, Massachusetts 02138, U.S.A. cohort of 41 337 gravidas and their offspring in the U.S.A. there was no evidence that aspirin taken in pregnancy is a cause of stillbirth, neonatal death, or reduced birth-weight. The women were divided into those who were not exposed to aspirin (14 956), those with intermediate exposure (24 866), and those who were heavily exposed (1515). Stillbirth-rates were similar for all three groups. Differences in neonatal death-rates and mean birth-weights were slight and none were statistically significant.
Summary
Trends
were
In
a
opposite
among
approximately equal
numbers of White and Black children. Introduction THE results of a recent study suggest that salicylates during pregnancy may reduce birth-weight and
taken
increase the risk of perinatal death. 12 We report vestigation into these issues.
an
in-
Methods The Collaborative Perinatal Project is described in the prethe purposes of the present study, the cohort of 50 282 mother-child pairs was modified as follows: (1) When a mother was enrolled in the study more than once, a random pregnancy was selected. This was done because perinatal deaths in prior siblings may increase the risk of subsequent perinatal death. (2) Pregnancies lasting less than 7 lunar months were excluded, since as explained below, the definition of heavy aspirin exposure used here was partly dependent upon the duration of pregnancy. The reduced cohort consisted of 41 337 mother-child pairs, among which there were 520 stillbirths (1.3%) and 437 neo-
ceding papery For
natal deaths (11%). 26 381 women used
aspirin at some time during pregnancy. defined, for pregnancies lasting at least "Heavy exposure" eight lunar months, as aspirin taken for at least eight days per lunar month in at least six lunar months. For pregnancies of seven lunar months duration, the drug had to be taken for at least eight days in each of at least five lunar months. There were 1515 heavily exposed mother-child pairs. Aspirin was less commonly used by non-smokers, nulliparz, was
White women, and normotensives. In addition, because of the way in which heavy exposure was defined in this study, such exposure was more common in pregnancies that lasted longer. Cigarette-smoking and parity were also related to perinatal mortality. Therefore perinatal mortality-rates were standardised by the direct method for these two factors. Birthweight was related to duration of pregnancy, cigarette-smoking, birth order, and pre-existing hypertension complicating pre-eclampsia. Mean birth-weights were adjusted by linear regression techniques for these factors. The standardised analyses were carried out separately for the White and Black ethnic groups. Because of small numbers, 3124 Puerto Rican mother-child pairs were included only when considering overall unadjusted perinatal mortality-rates (table I). In all further analyses they were excluded, leaving 38 213 mother-child pairs.
Results shows the stillbirth and neonatal death-rates to aspirin exposure during pregnancy. The rates were similar for each category of exposure. In table n, standardised stillbirth and neonatal deathrates are shown according to ethnic group. Among White children the stillbirth-rates were similar for all categories of exposure. Among Black children, the stillbirth-rates decreased with increasing antenatal exposure to aspirin, but this trend was not statistically significant. The neonatal death-rate was higher in White children exposed in utero to heavy aspirin-use than in non-exposed children. However, among heavily exposed Black children, the neonatal death-rate was lower. Neither of the differences was statistically significant. Other exposure to aspirin was not associated with neonatal death.
Table
i
according
TABLE I-STILLBIRTHS AND NEONATAL DEATHS IN RELATION TO ASPIRIN EXPOSURE DURING PREGNANCY
