- Email: [email protected]
Surveillance of developmental disabilities with an emphasis on special studies
Reproductive Toxicology, Vol. 11, Nos. 2/3, pp. 271-274, 1997 Copyright 0 1997 Elsevier Science Inc. Printed in the USA. All rights reserved 0890-6238/97 $17.00 + .OO ELSEVIER
PI1 SOS90-6238(96)00144-X
SURVEILLANCE
OF DEVELOPMENTAL DISABILITIES EMPHASIS ON SPECIAL STUDIES
WITH AN
COLEEN A. BOYLE Developmental Disabilities Branch, Division of Birth Defects and Developmental Disabilities, National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, Georgia 30333 Abstract - The developing central nervous system seems to be particularly vulnerable to chemical insults. A model for developmental disabilities surveillance is presented that provides a reasonable framework for monitoring the prevalence of various developmental abnormalities in human populations. Effective monitoring will not only increase the likelihood of detecting the adverse effects of new physical or chemical agents in the environment but will provide a readily available case series for specially directed case-control studies. A specific example is provided of a large case-control study of cerebral palsy and intrapartum magnesium exposure among very low birth weight children, which is being conducted within the framework of a developmental disabilities surveillance program. 0 1997 Elsevier Science Inc. Key Words: chemical
insults; Developmental
disabilities
surveillance
INTRODUCTION
which allows potentially more freely (5).
The developing central nervous system (CNS) seems to be particularly vulnerable to various chemical and physical agents. Important examples include the deleterious effects of in utero and early childhood exposure to alcohol (I), methyl mercury (2), polychlorinated biphenyls (3), and lead (4). Each of these has been shown to cause severe cognitive or motor impairment at high levels and to have more subtle effects at lower levels. Why does the developing nervous system seem particularly vulnerable? Part of the reason undoubtedly lies in its complexity. In contrast to other organ systems that develop over limited time periods, the CNS develops over an extended period of time from early in gestation until well into the first year of postnatal life, making it susceptible to injury over many developmental stages (5). Furthermore, the CNS produces many types of cells, and if some are lost, a whole category of cells may be destroyed. Other organ systems have fewer cell types and can, thus, more easily compensate for a similar insult. These two factors (i.e., the CNS’s extended period of development and the many types of cells it contains) result in a greater likelihood of injury to the CNS. Other factors that contribute to CNS vulnerability include the inability of the CNS to replace damaged cells even during development and the immaturity of the blood-brain barrier in the developing brain,
Address correspondence abilities Branch, Division of abilities, National Center for ease Control and Prevention,
damaging
substances
EPIDEMIOLOGIC APPROACHES STUDYING DEVELOPMENTAL ABNORMALITIES
to pass
TO
There are several epidemiologic approaches to studying the risk for developmental abnormalities from chemical exposures in human populations. The most common approach is a cohort study of an entire population (or subset of the population) of children followed from pregnancy or birth to early childhood to examine changes in neurologic, behavioral, and cognitive functioning associated with a particular chemical exposure. Often this approach is undertaken with a unique exposed population; however, it is generally time consuming and costly because researchers must periodically test the exposure of all children (measured through use of a biologic marker, if available) and their outcome (usually measured through standardized developmental and neurologic evaluations). Examples of studies in which this approach has been used include studies of children exposed in utero or postnatally to polychlorinated biphenyls (6) and methyl mercury (7) from maternal fish consumption, studies of children exposed to lead from ambient environmental sources (4), and general population studies of children exposed in utero to alcohol (1). An alternative epidemiologic approach to studying the risk for developmental abnormalities is to conduct ongoing surveillance of such outcomes. In and of itself, surveillance of developmental abnormalities will provide
to Coleen A. Boyle, Developmental DisBirth Defects and Developmental DisEnvironmental Health, Centers for DisMailstop F-15, Atlanta, GA 30333. 271
272
Reproductive Toxicology
important epidemiologic information about their prevalence in various subgroups of the population (8). The effective use of surveillance data can also enhance the likelihood of detecting the introduction into the environment of new physical or chemical agents that may increase (or decrease) the prevalence of developmental abnormalities. Prerequisites of effective monitoring (see Table 1) include having an adequate population size, monitoring population subgroups so as to identify those with maximum potential for exposure, and classifying outcomes into etiologically homogeneous subgroups for analysis (9). Ways to boost sample size include pooling data from several individual systems (which has been the model for cancer and birth defects surveillance systems) and using nationwide surveillance mechanisms, such as those available from the vital records birth and death files. To monitor population subgroups with maximum potential for exposure, researchers might, for example, focus on geographic areas where ambient environmental exposures are likely to occur (e.g., areas where the drinking water system is contaminated with bacteria and toxic chemicals following an environmental disaster, such as a flood or a hurricane) or, they might focus on ethnic subgroups with similar customs or practices (e.g., subsistence fish-eating populations with increased exposures to environmental chemicals). By focusing on outcomes that have distinct epidemiologic patterns, researchers can increase the etiologic homogeneity of the outcomes. For example, in our studies of mental retardation, we have found that mental retardation accompanied by other neurologic conditions (e.g., cerebral palsy, epilepsy, CNS birth defects) had epidemiologic characteristics very different from those of isolated mental retardation regardless of the severity of the mental retardation (IO). Thus, to detect factors that may influence mental retardation, we monitor these two categories separately. A MODEL FOR THE SURVEILLANCE DEVELOPMENTAL DISABILITIES
OF
In 1992, we developed a population-based surveillance program to monitor the occurrence of developmenTable I. Methods for improving the ability of a surveillance system to detect changes in the prevalence of disease associated with the introduction of a new exposure Method
Examples
I. Ensure that the system
Combine data from several
covers an adequate population size. 2. Monitor high-risk subgroups.
surveillance systems or use national vital record systems. Monitor geographic areas or ethnic populations with an increased likelihood for exposure. Group outcomes with distinct epidemiologic characteristics.
3. Classify outcomes into homogeneous subgroups.
Volume I I, Numbers 213, 1997
tal disabilities in the metropolitan Atlanta area (8). To do this, we took advantage of the fact that federal law requires public schools to provide educational services to children with developmental problems and, thus, to identify and maintain information concerning such children (11). Although we have found that public schools are the primary data sources for children with developmental disabilities, we have also identified as other surveillance sources, the sites where children with developmental disabilities are likely to be diagnosed or to receive services for their disability, such as pediatric hospitals and private and state programs for children with developmental disabilities (10). The objective of the surveillance system we developed is to ascertain all children within the five-county metropolitan Atlanta area who have one or more of four developmental disabilities-mental retardation, cerebral palsy, hearing impairment, vision impairment. Clearly, this system does not monitor all developmental disabilities; however, the system is flexible in that other developmental disabilities that fit the surveillance model can be added as resources allow. In fact, we are currently developing methods to add autism to the list of conditions monitored. The surveillance methodology for this system includes an ongoing review of records at all sources (i.e., schools, hospitals, and selected state and private programs for the developmentally disabled). Children who meet one or more of our surveillance definitions for a developmental disability are included as case subjects. A developmental pediatrician makes the final determination for questionable cases by reviewing all available records on the child involved. Detailed information is then collected from the medical, school, or other source records of children who meet the case defintion. This information includes the children’s sociodemographic characteristics and diagnostic information, such as the underlying etiology of their disability (e.g., chromosomal defect, motor vehicle-related head trauma) and any associated medical conditions they may have. This surveillance data set allows us to use census information to examine the current burden of developmental disabilities in the population. In addition, because we link with Georgia birth files for the subset of children born in Georgia, we can examine the risk for developmental disabilities associated with selected sociodemographic, maternal, and infant characteristics. To make the surveillance system as sensitive as possible to changes in the environment (i.e., the introduction of a new medical or environmental agent), we need to have etiologically homogeneous groups. To this end, we attempt to classify the developmental disabilities into meaningful subcategories. As mentioned above, the classification scheme for mental retardation is based on the presence of other
Surveillance
of developmental
neurologic conditions. In the case of cerebral palsy, a developmental pediatrician determines the type of cerebral palsy (e.g., spastic, dyskinetic, ataxic), the level of functioning of the child (ambulatory/nonambulatory), the associated neurologic conditions (e.g., other developmental disabilities, CNS birth defects) and the gestational age and birth weight of the child. By monitoring the prevalence of cerebral palsy with respect to these factors, we gain more sensitivity in detecting changes in particular subtypes of cerebral palsy. In addition, we minimize misclassification by following up on all questionable diagnoses to determine whether another diagnosis (e.g., a progressive neurologic disorder) may be more appropriate. The CDC developmental disability surveillance system is unique in that it is the only system of its kind in this country and the only one in the world that includes multiple developmental disabilities. SPECIAL
STUDIES
An important aspect of the surveillance methodology of this system is that it allows for the investigation of specific causal agents in a timely manner with better statistical power and potentially lower cost than would be possible with a cohort study. Let me give a current and important example. In February 1995, investigators at the California Department of Health Services and National Institutes of Health reported in a case
disabilities
0 C. A. BOYLE
273
gaps concerning some important exposure information, such as timing and dose. In addition, because the California study was an observational one, some researchers voiced methodologic concerns as to whether the association between magnesium sulfate administration and reduction in risk for cerebral palsy was an artifact of a clinical decision-making process that happened to be correlated with magnesium use. In other words, they wondered whether the decision to treat women with magnesium prior to their preterm delivery is a marker for a more favorable neurodevelopmental outcome of their children irrespective of magnesium use. The only way to determine whether such a bias is inherent in the clinical decision-making process is to conduct a clinical trial in which the selection of women to receive magnesium sulfate is experimentally controlled. However, a clinical trial of magnesium supplementation among women with eminent preterm delivery is feasible only if a very large base population is being monitored. This large population is necessary because 1) very low birth weight (i.e., < 1500 g) accounts for only 1% of births (15); 2) only 60-80% of very low birth weight babies survive (depending on their gestational age) (16); 3) even though very low birth weight survivors have a much higher risk for cerebral palsy, only 5% will develop the condition (as opposed to 0.2% of the general population) (13); and 4) only 50% of all women delivering preterm would be eligible for a clinical trial. Another complication in the design of a trial is that the diagnosis of cerebral palsy is difficult, and many transient effects of very low birth weight in very young children are misdiagnosed as cerebral palsy. Consequently, children should be followed for the development of cerebral palsy to at least the age of 2. Figure 1 shows that about 2000 children are needed in the experimental and control arms (combined) of a trial if the trial is to have 80% power to detect a 50% reduction in the rate of cerebral palsy among 2-year-old children that is due to magnesium exposure. Although 2000 seems like a reasonable sample size, to end up with this number of children at age 2, one would need an initial sample of approximately 10,000 pregnant women who have eminent early delivery from a base population of approximately 1 million pregnant women. Such an effort would clearly require a concerted multisite effort of long duration. This example is given not to suggest that clinical trials do not have an important part to play in affirming causality, but to emphasize their complexity and lack of timeliness. Our surveillance system allowed us to rapidly address the need for further study of the use of intrapartum magnesium exposure and cerebral palsy in two ways. The first was by linking data from a CDC cohort study of very low birth weight survivors (which had information
214
Reproductive
Toxicology
1% of all deliveries are of very low birth weight infants. 5% of very low birth weight survivors will develop cerebral palsy. Treatment reduces the rate of cerebral palsy by 50%.
Adjusted for 50% eligibility Adjusted for 50% participation Total number of deliveries monitored
that must be
1968 .t 2460 1 4920 1 9840 1 984,000
Fig. 1. Estimates of the total number of deliveries that must be monitored to conduct a clinical trial of the effect that prophylactic use of magnesium sulfate in very low birth weight infants on the prevention of cerebral palsy. on maternal drug use during the delivery admission) with data from the surveillance system to determine the prevalence of cerebral palsy among very low birth weight survivors whose mothers were and whose mothers were not administered magnesium sulfate. The results of this analysis support the findings of the California study for cerebral palsy and also suggest that magnesium sulfate may protect against other adverse developmental outcomes (17). Our second approach to studying the neurodevelopmental effects of intrapartum magnesium exposure was to do a large population-based case-control study using as case subjects all children who were born in 19811989, who weighed < 1750 g at birth, and who were identified by the surveillance system as having cerebral palsy. Using Georgia birth records, we selected a control group among metropolitan Atlanta infants who were born in the same years, weighed < 1750 g and survived to 1 year. We are now well under way in reviewing the mothers labor and delivery records and the children’s newborn records to reconstruct the administration of magnesium sulfate and other drugs to the mothers during the intrapartum period and to the infants during their first few weeks of life. This surveillance framework provides for a case
1 1, Numbers 2/3, 1997
toring of patterns of such outcomes in children, but if used effectively, they allow for surveillance of the adverse effects of agents newly introduced into the environment and for the rapid development of case+control studies designed to measure the effects of such agents.
Assumptions:
Number of participants (experimental + control) needed for 80% power at (Y = 0.05 Adjusted for 80% survival
Volume
systems of developmental abnormaliNot only do they allow for the moni-
REFERENCES 1. Streissguth AP, Barr HM, Sampson PD. Moderate prenatal alcohol exposure: effects on child IQ and learning problems at age 7-l/2 years. Alcohol Clin Exp Res. 1990; 14:662Z9. 2. Harada M. Congenital Minamata disease: intrauterine methyl mercury poisoning. Teratology. 1878: 18:285-S. 3. Yung-Cheng JC, Yue-Liang G, Chen-Chin H, et al. Cognitive development of Yu-Cheng (oil disease) children prenatally exposed to heat-graded PCBs. JAMA. 1992;268:32 13-8. 4. Baghurst PA, McMichael AJ, Wigg NR, et al. Environmental exposure to lead and children’s intelligence at the age of seven years: the Port Pirie Cohort Study. N Engl J Med. 1992;337: 127984. 5. Rodier PM. Vulnerable periods and processes during central nervous system development. Environ Health Perspect. 1994; 102: 1214. 6. Jacobson JL, Jacobson SW, Humphrey HEB. Effects of in utero exposure to polychlorinated biphenyls and related contaminants on cognitive functioning in young children. J Pediatr. lYYO;ll65:3845. 7. McKeown-Eyssen G, Ruedy J, Neims A. Methyl mercury exposure in northern Quebec. II. Neurologic findings in children. Am J Epidemiol. lY83;l l&470-9. 8. Boyle CA, Yeargin-Allsopp M, Doernberg N, Horngreen P, Murphy CC, Schendel D. Prevalance of selected developmental disabilities in children aged 3-10 Years: the Metropolitan Atlanta Developmental Disabilities Surveillance Program, 199 1. MMWR CDC Surveill Summ. 1996;45: I-. 9. Khoury MJ, Holtzman NA. On the ability of birth defects monitoring to detect new teratogens. Am J Epidemiol. 1987; 126: 136 43. 10. Drews CD, Yeargin-Allsopp M, Decoufle P, Murphy CC. Variation in the influence of selected socio-demographic risk factors for mental retardation. Am J Public Health. 1995;85:32Y-34. 11. The Individuals with Disabilities Education Act Amendments of 1991, PL 102.119. 20 USC 1400 et seq. Fed Reg. 1991;137: 1901. 12. Nelson KB, Grether JK. Can magnesium sulfate reduce the risk of cerebral palsy in very low birth weight infants? Pediatrics. 1995; 951263-9. 13. Kuban KCK, Leviton A. Cerebral palsy. N Engl J Med. 1994;330: 188-94. 14. Petterson B, Nelson KB, Watson L. Stanley F. Twins, triplets, and cerebral palsy in births in Western Australia in the 1980’s. Br J Med. 1993;307:123943. 15. National Center for Health Statistics. Advanced report of final natality statistics, 1985. Monthly Vital Statistics Report. Vol 36, No. 4 supp. (DHHS pub. no. 87-l 120). Hyattsville, MD: Public Health Service; 1987. 16. Copper RL, Goldenberg RL, Greasy RK, et al. A multicenter study of preterm birthweight and gestational age-specific neonatal mortality. Am J Obstet Gynecol. lYY3;168:78-84. 17. Schendel DE, Berg CJ, Yeargin-Allsopp M, Boyle CA, Decoufle P. Prenatal magnesium sulfate exposure and the risk for cerebral palsy or mental retardation among very low birth weight children aged 3-5 years. JAMA, in press.
PI1 SOS90-6238(96)00144-X
SURVEILLANCE
OF DEVELOPMENTAL DISABILITIES EMPHASIS ON SPECIAL STUDIES
WITH AN
COLEEN A. BOYLE Developmental Disabilities Branch, Division of Birth Defects and Developmental Disabilities, National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, Georgia 30333 Abstract - The developing central nervous system seems to be particularly vulnerable to chemical insults. A model for developmental disabilities surveillance is presented that provides a reasonable framework for monitoring the prevalence of various developmental abnormalities in human populations. Effective monitoring will not only increase the likelihood of detecting the adverse effects of new physical or chemical agents in the environment but will provide a readily available case series for specially directed case-control studies. A specific example is provided of a large case-control study of cerebral palsy and intrapartum magnesium exposure among very low birth weight children, which is being conducted within the framework of a developmental disabilities surveillance program. 0 1997 Elsevier Science Inc. Key Words: chemical
insults; Developmental
disabilities
surveillance
INTRODUCTION
which allows potentially more freely (5).
The developing central nervous system (CNS) seems to be particularly vulnerable to various chemical and physical agents. Important examples include the deleterious effects of in utero and early childhood exposure to alcohol (I), methyl mercury (2), polychlorinated biphenyls (3), and lead (4). Each of these has been shown to cause severe cognitive or motor impairment at high levels and to have more subtle effects at lower levels. Why does the developing nervous system seem particularly vulnerable? Part of the reason undoubtedly lies in its complexity. In contrast to other organ systems that develop over limited time periods, the CNS develops over an extended period of time from early in gestation until well into the first year of postnatal life, making it susceptible to injury over many developmental stages (5). Furthermore, the CNS produces many types of cells, and if some are lost, a whole category of cells may be destroyed. Other organ systems have fewer cell types and can, thus, more easily compensate for a similar insult. These two factors (i.e., the CNS’s extended period of development and the many types of cells it contains) result in a greater likelihood of injury to the CNS. Other factors that contribute to CNS vulnerability include the inability of the CNS to replace damaged cells even during development and the immaturity of the blood-brain barrier in the developing brain,
Address correspondence abilities Branch, Division of abilities, National Center for ease Control and Prevention,
damaging
substances
EPIDEMIOLOGIC APPROACHES STUDYING DEVELOPMENTAL ABNORMALITIES
to pass
TO
There are several epidemiologic approaches to studying the risk for developmental abnormalities from chemical exposures in human populations. The most common approach is a cohort study of an entire population (or subset of the population) of children followed from pregnancy or birth to early childhood to examine changes in neurologic, behavioral, and cognitive functioning associated with a particular chemical exposure. Often this approach is undertaken with a unique exposed population; however, it is generally time consuming and costly because researchers must periodically test the exposure of all children (measured through use of a biologic marker, if available) and their outcome (usually measured through standardized developmental and neurologic evaluations). Examples of studies in which this approach has been used include studies of children exposed in utero or postnatally to polychlorinated biphenyls (6) and methyl mercury (7) from maternal fish consumption, studies of children exposed to lead from ambient environmental sources (4), and general population studies of children exposed in utero to alcohol (1). An alternative epidemiologic approach to studying the risk for developmental abnormalities is to conduct ongoing surveillance of such outcomes. In and of itself, surveillance of developmental abnormalities will provide
to Coleen A. Boyle, Developmental DisBirth Defects and Developmental DisEnvironmental Health, Centers for DisMailstop F-15, Atlanta, GA 30333. 271
272
Reproductive Toxicology
important epidemiologic information about their prevalence in various subgroups of the population (8). The effective use of surveillance data can also enhance the likelihood of detecting the introduction into the environment of new physical or chemical agents that may increase (or decrease) the prevalence of developmental abnormalities. Prerequisites of effective monitoring (see Table 1) include having an adequate population size, monitoring population subgroups so as to identify those with maximum potential for exposure, and classifying outcomes into etiologically homogeneous subgroups for analysis (9). Ways to boost sample size include pooling data from several individual systems (which has been the model for cancer and birth defects surveillance systems) and using nationwide surveillance mechanisms, such as those available from the vital records birth and death files. To monitor population subgroups with maximum potential for exposure, researchers might, for example, focus on geographic areas where ambient environmental exposures are likely to occur (e.g., areas where the drinking water system is contaminated with bacteria and toxic chemicals following an environmental disaster, such as a flood or a hurricane) or, they might focus on ethnic subgroups with similar customs or practices (e.g., subsistence fish-eating populations with increased exposures to environmental chemicals). By focusing on outcomes that have distinct epidemiologic patterns, researchers can increase the etiologic homogeneity of the outcomes. For example, in our studies of mental retardation, we have found that mental retardation accompanied by other neurologic conditions (e.g., cerebral palsy, epilepsy, CNS birth defects) had epidemiologic characteristics very different from those of isolated mental retardation regardless of the severity of the mental retardation (IO). Thus, to detect factors that may influence mental retardation, we monitor these two categories separately. A MODEL FOR THE SURVEILLANCE DEVELOPMENTAL DISABILITIES
OF
In 1992, we developed a population-based surveillance program to monitor the occurrence of developmenTable I. Methods for improving the ability of a surveillance system to detect changes in the prevalence of disease associated with the introduction of a new exposure Method
Examples
I. Ensure that the system
Combine data from several
covers an adequate population size. 2. Monitor high-risk subgroups.
surveillance systems or use national vital record systems. Monitor geographic areas or ethnic populations with an increased likelihood for exposure. Group outcomes with distinct epidemiologic characteristics.
3. Classify outcomes into homogeneous subgroups.
Volume I I, Numbers 213, 1997
tal disabilities in the metropolitan Atlanta area (8). To do this, we took advantage of the fact that federal law requires public schools to provide educational services to children with developmental problems and, thus, to identify and maintain information concerning such children (11). Although we have found that public schools are the primary data sources for children with developmental disabilities, we have also identified as other surveillance sources, the sites where children with developmental disabilities are likely to be diagnosed or to receive services for their disability, such as pediatric hospitals and private and state programs for children with developmental disabilities (10). The objective of the surveillance system we developed is to ascertain all children within the five-county metropolitan Atlanta area who have one or more of four developmental disabilities-mental retardation, cerebral palsy, hearing impairment, vision impairment. Clearly, this system does not monitor all developmental disabilities; however, the system is flexible in that other developmental disabilities that fit the surveillance model can be added as resources allow. In fact, we are currently developing methods to add autism to the list of conditions monitored. The surveillance methodology for this system includes an ongoing review of records at all sources (i.e., schools, hospitals, and selected state and private programs for the developmentally disabled). Children who meet one or more of our surveillance definitions for a developmental disability are included as case subjects. A developmental pediatrician makes the final determination for questionable cases by reviewing all available records on the child involved. Detailed information is then collected from the medical, school, or other source records of children who meet the case defintion. This information includes the children’s sociodemographic characteristics and diagnostic information, such as the underlying etiology of their disability (e.g., chromosomal defect, motor vehicle-related head trauma) and any associated medical conditions they may have. This surveillance data set allows us to use census information to examine the current burden of developmental disabilities in the population. In addition, because we link with Georgia birth files for the subset of children born in Georgia, we can examine the risk for developmental disabilities associated with selected sociodemographic, maternal, and infant characteristics. To make the surveillance system as sensitive as possible to changes in the environment (i.e., the introduction of a new medical or environmental agent), we need to have etiologically homogeneous groups. To this end, we attempt to classify the developmental disabilities into meaningful subcategories. As mentioned above, the classification scheme for mental retardation is based on the presence of other
Surveillance
of developmental
neurologic conditions. In the case of cerebral palsy, a developmental pediatrician determines the type of cerebral palsy (e.g., spastic, dyskinetic, ataxic), the level of functioning of the child (ambulatory/nonambulatory), the associated neurologic conditions (e.g., other developmental disabilities, CNS birth defects) and the gestational age and birth weight of the child. By monitoring the prevalence of cerebral palsy with respect to these factors, we gain more sensitivity in detecting changes in particular subtypes of cerebral palsy. In addition, we minimize misclassification by following up on all questionable diagnoses to determine whether another diagnosis (e.g., a progressive neurologic disorder) may be more appropriate. The CDC developmental disability surveillance system is unique in that it is the only system of its kind in this country and the only one in the world that includes multiple developmental disabilities. SPECIAL
STUDIES
An important aspect of the surveillance methodology of this system is that it allows for the investigation of specific causal agents in a timely manner with better statistical power and potentially lower cost than would be possible with a cohort study. Let me give a current and important example. In February 1995, investigators at the California Department of Health Services and National Institutes of Health reported in a case
disabilities
0 C. A. BOYLE
273
gaps concerning some important exposure information, such as timing and dose. In addition, because the California study was an observational one, some researchers voiced methodologic concerns as to whether the association between magnesium sulfate administration and reduction in risk for cerebral palsy was an artifact of a clinical decision-making process that happened to be correlated with magnesium use. In other words, they wondered whether the decision to treat women with magnesium prior to their preterm delivery is a marker for a more favorable neurodevelopmental outcome of their children irrespective of magnesium use. The only way to determine whether such a bias is inherent in the clinical decision-making process is to conduct a clinical trial in which the selection of women to receive magnesium sulfate is experimentally controlled. However, a clinical trial of magnesium supplementation among women with eminent preterm delivery is feasible only if a very large base population is being monitored. This large population is necessary because 1) very low birth weight (i.e., < 1500 g) accounts for only 1% of births (15); 2) only 60-80% of very low birth weight babies survive (depending on their gestational age) (16); 3) even though very low birth weight survivors have a much higher risk for cerebral palsy, only 5% will develop the condition (as opposed to 0.2% of the general population) (13); and 4) only 50% of all women delivering preterm would be eligible for a clinical trial. Another complication in the design of a trial is that the diagnosis of cerebral palsy is difficult, and many transient effects of very low birth weight in very young children are misdiagnosed as cerebral palsy. Consequently, children should be followed for the development of cerebral palsy to at least the age of 2. Figure 1 shows that about 2000 children are needed in the experimental and control arms (combined) of a trial if the trial is to have 80% power to detect a 50% reduction in the rate of cerebral palsy among 2-year-old children that is due to magnesium exposure. Although 2000 seems like a reasonable sample size, to end up with this number of children at age 2, one would need an initial sample of approximately 10,000 pregnant women who have eminent early delivery from a base population of approximately 1 million pregnant women. Such an effort would clearly require a concerted multisite effort of long duration. This example is given not to suggest that clinical trials do not have an important part to play in affirming causality, but to emphasize their complexity and lack of timeliness. Our surveillance system allowed us to rapidly address the need for further study of the use of intrapartum magnesium exposure and cerebral palsy in two ways. The first was by linking data from a CDC cohort study of very low birth weight survivors (which had information
214
Reproductive
Toxicology
1% of all deliveries are of very low birth weight infants. 5% of very low birth weight survivors will develop cerebral palsy. Treatment reduces the rate of cerebral palsy by 50%.
Adjusted for 50% eligibility Adjusted for 50% participation Total number of deliveries monitored
that must be
1968 .t 2460 1 4920 1 9840 1 984,000
Fig. 1. Estimates of the total number of deliveries that must be monitored to conduct a clinical trial of the effect that prophylactic use of magnesium sulfate in very low birth weight infants on the prevention of cerebral palsy. on maternal drug use during the delivery admission) with data from the surveillance system to determine the prevalence of cerebral palsy among very low birth weight survivors whose mothers were and whose mothers were not administered magnesium sulfate. The results of this analysis support the findings of the California study for cerebral palsy and also suggest that magnesium sulfate may protect against other adverse developmental outcomes (17). Our second approach to studying the neurodevelopmental effects of intrapartum magnesium exposure was to do a large population-based case-control study using as case subjects all children who were born in 19811989, who weighed < 1750 g at birth, and who were identified by the surveillance system as having cerebral palsy. Using Georgia birth records, we selected a control group among metropolitan Atlanta infants who were born in the same years, weighed < 1750 g and survived to 1 year. We are now well under way in reviewing the mothers labor and delivery records and the children’s newborn records to reconstruct the administration of magnesium sulfate and other drugs to the mothers during the intrapartum period and to the infants during their first few weeks of life. This surveillance framework provides for a case
1 1, Numbers 2/3, 1997
toring of patterns of such outcomes in children, but if used effectively, they allow for surveillance of the adverse effects of agents newly introduced into the environment and for the rapid development of case+control studies designed to measure the effects of such agents.
Assumptions:
Number of participants (experimental + control) needed for 80% power at (Y = 0.05 Adjusted for 80% survival
Volume
systems of developmental abnormaliNot only do they allow for the moni-
REFERENCES 1. Streissguth AP, Barr HM, Sampson PD. Moderate prenatal alcohol exposure: effects on child IQ and learning problems at age 7-l/2 years. Alcohol Clin Exp Res. 1990; 14:662Z9. 2. Harada M. Congenital Minamata disease: intrauterine methyl mercury poisoning. Teratology. 1878: 18:285-S. 3. Yung-Cheng JC, Yue-Liang G, Chen-Chin H, et al. Cognitive development of Yu-Cheng (oil disease) children prenatally exposed to heat-graded PCBs. JAMA. 1992;268:32 13-8. 4. Baghurst PA, McMichael AJ, Wigg NR, et al. Environmental exposure to lead and children’s intelligence at the age of seven years: the Port Pirie Cohort Study. N Engl J Med. 1992;337: 127984. 5. Rodier PM. Vulnerable periods and processes during central nervous system development. Environ Health Perspect. 1994; 102: 1214. 6. Jacobson JL, Jacobson SW, Humphrey HEB. Effects of in utero exposure to polychlorinated biphenyls and related contaminants on cognitive functioning in young children. J Pediatr. lYYO;ll65:3845. 7. McKeown-Eyssen G, Ruedy J, Neims A. Methyl mercury exposure in northern Quebec. II. Neurologic findings in children. Am J Epidemiol. lY83;l l&470-9. 8. Boyle CA, Yeargin-Allsopp M, Doernberg N, Horngreen P, Murphy CC, Schendel D. Prevalance of selected developmental disabilities in children aged 3-10 Years: the Metropolitan Atlanta Developmental Disabilities Surveillance Program, 199 1. MMWR CDC Surveill Summ. 1996;45: I-. 9. Khoury MJ, Holtzman NA. On the ability of birth defects monitoring to detect new teratogens. Am J Epidemiol. 1987; 126: 136 43. 10. Drews CD, Yeargin-Allsopp M, Decoufle P, Murphy CC. Variation in the influence of selected socio-demographic risk factors for mental retardation. Am J Public Health. 1995;85:32Y-34. 11. The Individuals with Disabilities Education Act Amendments of 1991, PL 102.119. 20 USC 1400 et seq. Fed Reg. 1991;137: 1901. 12. Nelson KB, Grether JK. Can magnesium sulfate reduce the risk of cerebral palsy in very low birth weight infants? Pediatrics. 1995; 951263-9. 13. Kuban KCK, Leviton A. Cerebral palsy. N Engl J Med. 1994;330: 188-94. 14. Petterson B, Nelson KB, Watson L. Stanley F. Twins, triplets, and cerebral palsy in births in Western Australia in the 1980’s. Br J Med. 1993;307:123943. 15. National Center for Health Statistics. Advanced report of final natality statistics, 1985. Monthly Vital Statistics Report. Vol 36, No. 4 supp. (DHHS pub. no. 87-l 120). Hyattsville, MD: Public Health Service; 1987. 16. Copper RL, Goldenberg RL, Greasy RK, et al. A multicenter study of preterm birthweight and gestational age-specific neonatal mortality. Am J Obstet Gynecol. lYY3;168:78-84. 17. Schendel DE, Berg CJ, Yeargin-Allsopp M, Boyle CA, Decoufle P. Prenatal magnesium sulfate exposure and the risk for cerebral palsy or mental retardation among very low birth weight children aged 3-5 years. JAMA, in press.