DEVELOPMENTAL MEDICINE & CHILD NEUROLOGY
ORIGINAL ARTICLE
Fever during pregnancy and motor development in children: a study within the Danish National Birth Cohort CHARLOTTE HOLST 1 * | SANNE ELLEGAARD JØRGENSEN 1 * | JAN WOHLFAHRT 1 | ANNE-MARIE NYBO ANDERSEN 2 | MADS MELBYE 1 1 Department of Epidemiology Research, Statens Serum Institut, Copenhagen; 2 Section of Social Medicine, Department of Public Health, University of Copenhagen, Copenhagen, Denmark. Correspondence to Mads Melbye at Department of Epidemiology Research, Statens Serum Institut, Artillerivej 5, DK-2300 Copenhagen, Denmark. E-mail: [email protected] *These authors contributed equally to this work. This article is commented on by Hashmi on pages 698–699 of this issue.
PUBLICATION DATA
Accepted for publication 2nd February 2015. Published online 20th March 2015. ABBREVIATIONS
DCD DCDQ’07
Developmental coordination disorder Developmental Coordination Disorder Questionnaire 2007
AIM The aim of this study was to examine how fever during pregnancy is associated with motor development in the child. METHOD This cohort study was based on data from females and their children, from the Danish National Birth Cohort, who took part in an 18-month and/or 7-year follow-up study. Information regarding fever (number of episodes, temperature, duration, and pregnancy week) was obtained around gestation week 12 and at the end of pregnancy. Assessments of motor development in early childhood were based on the ages at which the motor milestones ‘sitting unsupported’ (n=44 256) and ‘walking unassisted’ (n=53 959) were attained. The Developmental Coordination Disorder Questionnaire 2007 (DCDQ’07) was used to identify children with indication of developmental coordination disorder (DCD) at age 7 years (n=29 401). Any associations between the exposure to fever during pregnancy and motor development were estimated using Cox regression and logistic regression analyses. RESULTS Fever during pregnancy was reported by 15 234 (28.0%) participants in the 18month follow-up and by 7965 (26.9%) participants in the 7-year follow-up. Adjusted analyses showed no association between prenatal exposure to fever and either ‘sitting unsupported’ or ‘walking unassisted’. The proportion of children with indication of DCD was 3.1%. The odds ratio of indication of DCD if children were exposed to fever in utero was 1.29 (95% CI 1.12–1.49). However, no dose-response association was found. INTERPRETATION We found a significant association between maternal fever during pregnancy and DCD in children at age 7 years. The lack of a dose–response association might suggest that this association is explained by the underlying causes of the fever.
Fever during pregnancy is suspected to increase the risk of several adverse outcomes in the developing fetus.1 In human studies, associations have been reported between prenatal fever exposure and neural tube defects,1,2 congenital malformations including oral clefts,3 and heart defects.4 A recent systematic review provides support for an adverse health impact on children after exposure to fever in utero; however, the authors conclude that the evidence of associations is insufficient for several of the outcomes addressed, and more cohort studies are needed.5 In addition, results from animal studies have shown that exposure to hyperthermia in utero produces adverse outcomes in offspring. In 1967, Edwards6 reported an increased incidence of fetal resorption and congenital malformations in guinea pigs exposed to hyperthermia during fetal life. Since then, animal studies have shown that hyperthermia is associated with fetal death, cardiovascular malformations, and a delayed myogenesis.7–9 The development of the central © 2015 Mac Keith Press
nervous system appears to be particularly sensitive to elevated body temperatures,7,10,11 and hyperthermia has been associated with the disruption of the production and migration of neuronal precursor cells and, thus, may result in abnormal brain development.10 A child’s motor abilities are associated with the development of cognitive, emotional, and social skills, and are therefore essential for the overall well-being and development of a child.12 Most infants follow the same sequence of attainment of the motor milestones, but the ages at which they are attained vary. A pronounced delay or failure in achieving motor milestones may be a sign of a disturbance in a child’s motor development.13 Non-age-appropriate motor development in children which is not caused by a general medical condition is often referred to as developmental coordination disorder (DCD).14 The mechanisms underlying mild motor disorders are still unknown;15,16 however, intrauterine growth restriction, DOI: 10.1111/dmcn.12743
725
stress during pregnancy,17 and smoking during pregnancy18 are all prenatal factors which have been associated with impaired motor development in children. The aim of this study was to examine the relationship between prenatal fever exposure and motor development during childhood.
METHOD Study participants Study participants were recruited from the Danish National Birth Cohort, which is an ongoing nationwide cohort of pregnant females and their children. Between 1996 and 2002, 100 418 participants were recruited into the Danish National Birth Cohort. Recruitment took place at the first prenatal visit to the general practitioner, an examination in the first trimester that is offered to all pregnant females in Denmark. Sixty-four per cent of all general practitioners participated in the recruitment process. Inclusion criteria for the cohort were desire to carry the pregnancy to term and ability to speak Danish well enough to take part in telephone interviews. Approximately 60% of all eligible pregnant females chose to participate in the cohort, corresponding to approximately 30% of all births in Denmark during the enrolment period. Information on parental lifestyle and health was collected through four standardized computer-assisted telephone interviews around gestation week 12, at the end of the pregnancy, and approximately 6 months and 18 months after delivery. When the children were 7 years old, the primary caregiver completed a follow-up questionnaire about the health and development of the child. In this study, participants were restricted to those who gave birth to live singletons and who had answered the questions related to the occurrence of fever in both interviews conducted during the pregnancy (n=73 657). Furthermore, only the first enrolled pregnancy from each participant was included (Figure S1, supporting information online). Measurements Information about fever exposure was obtained from two computer-assisted telephone interviews conducted during pregnancy. The participants were asked about the occurrence of fever during pregnancy and, for each fever episode, they were asked about its duration, their maximum body temperature, and the pregnancy week in which the fever occurred. Participants who reported fever in at least one interview were categorized as ‘exposed’. The number of fever episodes (0, 1, ≥2, or unknown) was based on the maximum number of incidents reported in the first and second interviews. The highest reported temperature was categorized as below 39.0°C, 39.0°C or more, or unknown. The duration of the fever (1d, 2d, 3d, ≥4d, or unknown) was based on the longest episode reported. Furthermore, the total duration of all reported febrile episodes was calculated and categorized into similar groups. Information about pregnancy week for each reported fever episode was obtained from the first pregnancy interview, if the fever episode occurred before 726 Developmental Medicine & Child Neurology 2015, 57: 725–732
• • •
What this paper adds First study to investigate maternal fever during pregnancy in relation to motor development in the child. Demonstrates that maternal fever during pregnancy is associated with developmental coordination disorder at 7 years of age. The lack of a dose–response relationship suggests that this association is caused by the underlying causes of the fever.
the first interview, or from the second interview, if the fever episode occurred after the first interview. The timing of each fever episode was categorized into one of seven gestational periods: pregnancy weeks 1 to 4, 5 to 8, 9 to 12, 13 to 18, 19 to 24, 25 to 30, or 31 to 40. All fever episodes were categorized according to this timing variable. Information about a child’s motor development was obtained by interview approximately 18 months after birth, and from the 7-year questionnaire. At the 18-month postpartum interview, caregivers were asked to report the ages (in months and weeks) at which the child could first sit unsupported and walk unassisted. Complete data about the motor milestone ‘sitting unsupported’ and relevant covariates were available for 50 680 respondents. A total of 53 959 respondents had information about the age at ‘walking unassisted’ and relevant covariates. Information about the child’s motor development at 7 years of age was based on answers to the Developmental Coordination Disorder Questionnaire 2007 (DCDQ’07). The DCDQ’07 was included in the 7-year questionnaire of the Danish National Birth Cohort in February 2007; thus, questionnaires completed before this date did not include the answers to the DCDQ’07. The DCDQ’07 is a parent-administered questionnaire designed to screen for indication of DCD. The questionnaire consists of 15 items concerning control during movement, fine motor skills/ handwriting, and general coordination. Caregivers are asked to compare the child’s motor development to peers of the same sex using a 5-point Likert scale with the following answer categories: ‘Not true’, ‘A little true’, ‘Fairly true’, ‘True’, or ‘Very true’. The scores for each item (1– 5 points) are summed to give a total score (15–75 points). According to the DCDQ’07 manual, children aged 5 to 7 with a total score of 46 or below are classified as ‘indication of possible, or suspect for DCD’, versus ‘probably not DCD’.19 A total of 29 401 respondents completed all items in the DCDQ’07 and had information about relevant confounding factors.
Confounders Confounders were identified in two steps. Potential confounding factors were identified a priori based on literature and included maternal age, smoking status during pregnancy, occupational status, and number of older children in the household. Furthermore, we used the ‘change in estimate’ method to investigate the possible confounding effect of factors where the evidence was less solid. We considered a change of at least 10% as a basis for inclusion. Potential confounding factors that were investigated by the
‘change in estimate’ method were stress during pregnancy and age compared with older children in the household. None of these variables changed the effect estimates by 10% or more. Based on the above considerations, the associations were adjusted for maternal age at birth, maternal smoking status during pregnancy, number of older children in the household, and occupational status. Occupational status was based on information from the first interview, at which females were asked about their current or most recent occupation (within the last 6mo) and about the occupation of the child’s father. This information was coded according to the Danish version of the International Standard Classification of Occupation, which contains 372 job titles.20 Based on job titles, parents were both categorized as either ‘higher non-manual worker’, ‘lower nonmanual worker’, ‘skilled manual worker’, ‘student’, ‘unskilled manual-worker’, ‘not working’, or ‘occupation unknown’. The child’s socioeconomic status was categorized according to the highest ranking parent. The analyses between fever in a given pregnancy period and motor development were adjusted for the occurrence of fever in other pregnancy periods.
Statistical analysis Stratified Cox regression analysis was used to compare the ages at which the developmental milestones ‘sitting unsupported’ and ‘walking unassisted’ were achieved in children exposed to fever prenatally and children not exposed to fever. The time from birth to the age at which the child achieved these two milestones represented the ‘time to event’. Children who were not able to sit without support (n=49) or walk unassisted (n=606) at the time of the interview were censored at the time of the interview in the analysis. By testing for interaction between each independent variable and the log of survival time, the assumption of proportional hazard was evaluated. In the analysis of the milestone ‘sitting unsupported’, the assumption of proportional hazards was met for the number of febrile episodes, temperature, duration, and timing. However, the assumption was not met for the occurrence of fever. Upon closer examination of the data, we found that the hazard functions were proportional in the time period from the age of 5 months and onwards. Children with an age of less than 5 months at attainment of sitting unsupported were, therefore, not included in the analysis of the association between prenatal fever exposure and the age of attaining this milestone. This reduced the study population to 44 256. In addition, analyses were stratified with respect to confounders in a Cox regression analysis, as these did not meet the assumption of proportional hazards either. For analysis of age at ‘walking unassisted’, the assumption of proportional hazards was satisfied for all exposures, but not for the selected confounders. These analyses were, therefore, also stratified with respect to the confounders in a Cox regression analysis. To estimate the proportion of children sitting and walking unsupported at a given age, the Kaplan–Meier estimator was used.
Multiple logistic regression analysis was used to compare the risk of DCD among children prenatally exposed to fever with unexposed children. Analyses were adjusted for the confounders mentioned above. The significance level was set at a p value of 0.05. Analyses were performed using SAS software version 9.3 (SAS Institute Inc., Cary, NC, USA). All study participants gave their informed consent to the research, and the study and the publication of its results were approved by the Danish Data Protection Agency.
RESULTS In total, 15 234 (28.0%) of those who participated in the 18-month interview reported at least one episode of fever during pregnancy and, among these respondents, 11.3% reported two or more episodes of fever. Of the participants who reported at least one fever episode, 7918 (52.0%) gave specific information about their temperature, 13 559 (89.0%) specified the duration of the fever episode, and 12 599 (82.7%) reported the pregnancy week of at least one fever episode. In total, 7965 (26.9%) of those who participated in the 7-year interview reported fever during pregnancy and, of these participants, 860 (10.8%) had at least two episodes of fever, 4085 (51.3%) reported their specific temperature, while 7125 (89.5%) respondents provided information about the duration of the fever episode (s). Information about the timing of at least one fever episode was available for 6659 (83.6%) respondents. Table I shows the distributions of selected characteristics of mothers and children, grouped according to their exposure to fever during pregnancy. The mean age of the children at completion of the 18month interview was 19.1 months (range 17.4–34.4mo). The median age at attaining the early motor milestones was 6.5 months (range 4.5–18.0mo) for ‘sitting unsupported’ and 12.5 months (range 1.0–21.0mo) for ‘walking unassisted’. Adjusted analyses showed no association between prenatal fever exposure and sitting unsupported (hazard ratio 0.99, 95% CI 0.97–1.01) or prenatal fever exposure and walking unassisted (hazard ratio 1.01, 95% CI 0.99–1.03), as shown in Table II. Sensitivity analyses were performed in relation to outliers of the outcomes. The exclusion of implausible low values (sitting at age
ORIGINAL ARTICLE
Fever during pregnancy and motor development in children: a study within the Danish National Birth Cohort CHARLOTTE HOLST 1 * | SANNE ELLEGAARD JØRGENSEN 1 * | JAN WOHLFAHRT 1 | ANNE-MARIE NYBO ANDERSEN 2 | MADS MELBYE 1 1 Department of Epidemiology Research, Statens Serum Institut, Copenhagen; 2 Section of Social Medicine, Department of Public Health, University of Copenhagen, Copenhagen, Denmark. Correspondence to Mads Melbye at Department of Epidemiology Research, Statens Serum Institut, Artillerivej 5, DK-2300 Copenhagen, Denmark. E-mail: [email protected] *These authors contributed equally to this work. This article is commented on by Hashmi on pages 698–699 of this issue.
PUBLICATION DATA
Accepted for publication 2nd February 2015. Published online 20th March 2015. ABBREVIATIONS
DCD DCDQ’07
Developmental coordination disorder Developmental Coordination Disorder Questionnaire 2007
AIM The aim of this study was to examine how fever during pregnancy is associated with motor development in the child. METHOD This cohort study was based on data from females and their children, from the Danish National Birth Cohort, who took part in an 18-month and/or 7-year follow-up study. Information regarding fever (number of episodes, temperature, duration, and pregnancy week) was obtained around gestation week 12 and at the end of pregnancy. Assessments of motor development in early childhood were based on the ages at which the motor milestones ‘sitting unsupported’ (n=44 256) and ‘walking unassisted’ (n=53 959) were attained. The Developmental Coordination Disorder Questionnaire 2007 (DCDQ’07) was used to identify children with indication of developmental coordination disorder (DCD) at age 7 years (n=29 401). Any associations between the exposure to fever during pregnancy and motor development were estimated using Cox regression and logistic regression analyses. RESULTS Fever during pregnancy was reported by 15 234 (28.0%) participants in the 18month follow-up and by 7965 (26.9%) participants in the 7-year follow-up. Adjusted analyses showed no association between prenatal exposure to fever and either ‘sitting unsupported’ or ‘walking unassisted’. The proportion of children with indication of DCD was 3.1%. The odds ratio of indication of DCD if children were exposed to fever in utero was 1.29 (95% CI 1.12–1.49). However, no dose-response association was found. INTERPRETATION We found a significant association between maternal fever during pregnancy and DCD in children at age 7 years. The lack of a dose–response association might suggest that this association is explained by the underlying causes of the fever.
Fever during pregnancy is suspected to increase the risk of several adverse outcomes in the developing fetus.1 In human studies, associations have been reported between prenatal fever exposure and neural tube defects,1,2 congenital malformations including oral clefts,3 and heart defects.4 A recent systematic review provides support for an adverse health impact on children after exposure to fever in utero; however, the authors conclude that the evidence of associations is insufficient for several of the outcomes addressed, and more cohort studies are needed.5 In addition, results from animal studies have shown that exposure to hyperthermia in utero produces adverse outcomes in offspring. In 1967, Edwards6 reported an increased incidence of fetal resorption and congenital malformations in guinea pigs exposed to hyperthermia during fetal life. Since then, animal studies have shown that hyperthermia is associated with fetal death, cardiovascular malformations, and a delayed myogenesis.7–9 The development of the central © 2015 Mac Keith Press
nervous system appears to be particularly sensitive to elevated body temperatures,7,10,11 and hyperthermia has been associated with the disruption of the production and migration of neuronal precursor cells and, thus, may result in abnormal brain development.10 A child’s motor abilities are associated with the development of cognitive, emotional, and social skills, and are therefore essential for the overall well-being and development of a child.12 Most infants follow the same sequence of attainment of the motor milestones, but the ages at which they are attained vary. A pronounced delay or failure in achieving motor milestones may be a sign of a disturbance in a child’s motor development.13 Non-age-appropriate motor development in children which is not caused by a general medical condition is often referred to as developmental coordination disorder (DCD).14 The mechanisms underlying mild motor disorders are still unknown;15,16 however, intrauterine growth restriction, DOI: 10.1111/dmcn.12743
725
stress during pregnancy,17 and smoking during pregnancy18 are all prenatal factors which have been associated with impaired motor development in children. The aim of this study was to examine the relationship between prenatal fever exposure and motor development during childhood.
METHOD Study participants Study participants were recruited from the Danish National Birth Cohort, which is an ongoing nationwide cohort of pregnant females and their children. Between 1996 and 2002, 100 418 participants were recruited into the Danish National Birth Cohort. Recruitment took place at the first prenatal visit to the general practitioner, an examination in the first trimester that is offered to all pregnant females in Denmark. Sixty-four per cent of all general practitioners participated in the recruitment process. Inclusion criteria for the cohort were desire to carry the pregnancy to term and ability to speak Danish well enough to take part in telephone interviews. Approximately 60% of all eligible pregnant females chose to participate in the cohort, corresponding to approximately 30% of all births in Denmark during the enrolment period. Information on parental lifestyle and health was collected through four standardized computer-assisted telephone interviews around gestation week 12, at the end of the pregnancy, and approximately 6 months and 18 months after delivery. When the children were 7 years old, the primary caregiver completed a follow-up questionnaire about the health and development of the child. In this study, participants were restricted to those who gave birth to live singletons and who had answered the questions related to the occurrence of fever in both interviews conducted during the pregnancy (n=73 657). Furthermore, only the first enrolled pregnancy from each participant was included (Figure S1, supporting information online). Measurements Information about fever exposure was obtained from two computer-assisted telephone interviews conducted during pregnancy. The participants were asked about the occurrence of fever during pregnancy and, for each fever episode, they were asked about its duration, their maximum body temperature, and the pregnancy week in which the fever occurred. Participants who reported fever in at least one interview were categorized as ‘exposed’. The number of fever episodes (0, 1, ≥2, or unknown) was based on the maximum number of incidents reported in the first and second interviews. The highest reported temperature was categorized as below 39.0°C, 39.0°C or more, or unknown. The duration of the fever (1d, 2d, 3d, ≥4d, or unknown) was based on the longest episode reported. Furthermore, the total duration of all reported febrile episodes was calculated and categorized into similar groups. Information about pregnancy week for each reported fever episode was obtained from the first pregnancy interview, if the fever episode occurred before 726 Developmental Medicine & Child Neurology 2015, 57: 725–732
• • •
What this paper adds First study to investigate maternal fever during pregnancy in relation to motor development in the child. Demonstrates that maternal fever during pregnancy is associated with developmental coordination disorder at 7 years of age. The lack of a dose–response relationship suggests that this association is caused by the underlying causes of the fever.
the first interview, or from the second interview, if the fever episode occurred after the first interview. The timing of each fever episode was categorized into one of seven gestational periods: pregnancy weeks 1 to 4, 5 to 8, 9 to 12, 13 to 18, 19 to 24, 25 to 30, or 31 to 40. All fever episodes were categorized according to this timing variable. Information about a child’s motor development was obtained by interview approximately 18 months after birth, and from the 7-year questionnaire. At the 18-month postpartum interview, caregivers were asked to report the ages (in months and weeks) at which the child could first sit unsupported and walk unassisted. Complete data about the motor milestone ‘sitting unsupported’ and relevant covariates were available for 50 680 respondents. A total of 53 959 respondents had information about the age at ‘walking unassisted’ and relevant covariates. Information about the child’s motor development at 7 years of age was based on answers to the Developmental Coordination Disorder Questionnaire 2007 (DCDQ’07). The DCDQ’07 was included in the 7-year questionnaire of the Danish National Birth Cohort in February 2007; thus, questionnaires completed before this date did not include the answers to the DCDQ’07. The DCDQ’07 is a parent-administered questionnaire designed to screen for indication of DCD. The questionnaire consists of 15 items concerning control during movement, fine motor skills/ handwriting, and general coordination. Caregivers are asked to compare the child’s motor development to peers of the same sex using a 5-point Likert scale with the following answer categories: ‘Not true’, ‘A little true’, ‘Fairly true’, ‘True’, or ‘Very true’. The scores for each item (1– 5 points) are summed to give a total score (15–75 points). According to the DCDQ’07 manual, children aged 5 to 7 with a total score of 46 or below are classified as ‘indication of possible, or suspect for DCD’, versus ‘probably not DCD’.19 A total of 29 401 respondents completed all items in the DCDQ’07 and had information about relevant confounding factors.
Confounders Confounders were identified in two steps. Potential confounding factors were identified a priori based on literature and included maternal age, smoking status during pregnancy, occupational status, and number of older children in the household. Furthermore, we used the ‘change in estimate’ method to investigate the possible confounding effect of factors where the evidence was less solid. We considered a change of at least 10% as a basis for inclusion. Potential confounding factors that were investigated by the
‘change in estimate’ method were stress during pregnancy and age compared with older children in the household. None of these variables changed the effect estimates by 10% or more. Based on the above considerations, the associations were adjusted for maternal age at birth, maternal smoking status during pregnancy, number of older children in the household, and occupational status. Occupational status was based on information from the first interview, at which females were asked about their current or most recent occupation (within the last 6mo) and about the occupation of the child’s father. This information was coded according to the Danish version of the International Standard Classification of Occupation, which contains 372 job titles.20 Based on job titles, parents were both categorized as either ‘higher non-manual worker’, ‘lower nonmanual worker’, ‘skilled manual worker’, ‘student’, ‘unskilled manual-worker’, ‘not working’, or ‘occupation unknown’. The child’s socioeconomic status was categorized according to the highest ranking parent. The analyses between fever in a given pregnancy period and motor development were adjusted for the occurrence of fever in other pregnancy periods.
Statistical analysis Stratified Cox regression analysis was used to compare the ages at which the developmental milestones ‘sitting unsupported’ and ‘walking unassisted’ were achieved in children exposed to fever prenatally and children not exposed to fever. The time from birth to the age at which the child achieved these two milestones represented the ‘time to event’. Children who were not able to sit without support (n=49) or walk unassisted (n=606) at the time of the interview were censored at the time of the interview in the analysis. By testing for interaction between each independent variable and the log of survival time, the assumption of proportional hazard was evaluated. In the analysis of the milestone ‘sitting unsupported’, the assumption of proportional hazards was met for the number of febrile episodes, temperature, duration, and timing. However, the assumption was not met for the occurrence of fever. Upon closer examination of the data, we found that the hazard functions were proportional in the time period from the age of 5 months and onwards. Children with an age of less than 5 months at attainment of sitting unsupported were, therefore, not included in the analysis of the association between prenatal fever exposure and the age of attaining this milestone. This reduced the study population to 44 256. In addition, analyses were stratified with respect to confounders in a Cox regression analysis, as these did not meet the assumption of proportional hazards either. For analysis of age at ‘walking unassisted’, the assumption of proportional hazards was satisfied for all exposures, but not for the selected confounders. These analyses were, therefore, also stratified with respect to the confounders in a Cox regression analysis. To estimate the proportion of children sitting and walking unsupported at a given age, the Kaplan–Meier estimator was used.
Multiple logistic regression analysis was used to compare the risk of DCD among children prenatally exposed to fever with unexposed children. Analyses were adjusted for the confounders mentioned above. The significance level was set at a p value of 0.05. Analyses were performed using SAS software version 9.3 (SAS Institute Inc., Cary, NC, USA). All study participants gave their informed consent to the research, and the study and the publication of its results were approved by the Danish Data Protection Agency.
RESULTS In total, 15 234 (28.0%) of those who participated in the 18-month interview reported at least one episode of fever during pregnancy and, among these respondents, 11.3% reported two or more episodes of fever. Of the participants who reported at least one fever episode, 7918 (52.0%) gave specific information about their temperature, 13 559 (89.0%) specified the duration of the fever episode, and 12 599 (82.7%) reported the pregnancy week of at least one fever episode. In total, 7965 (26.9%) of those who participated in the 7-year interview reported fever during pregnancy and, of these participants, 860 (10.8%) had at least two episodes of fever, 4085 (51.3%) reported their specific temperature, while 7125 (89.5%) respondents provided information about the duration of the fever episode (s). Information about the timing of at least one fever episode was available for 6659 (83.6%) respondents. Table I shows the distributions of selected characteristics of mothers and children, grouped according to their exposure to fever during pregnancy. The mean age of the children at completion of the 18month interview was 19.1 months (range 17.4–34.4mo). The median age at attaining the early motor milestones was 6.5 months (range 4.5–18.0mo) for ‘sitting unsupported’ and 12.5 months (range 1.0–21.0mo) for ‘walking unassisted’. Adjusted analyses showed no association between prenatal fever exposure and sitting unsupported (hazard ratio 0.99, 95% CI 0.97–1.01) or prenatal fever exposure and walking unassisted (hazard ratio 1.01, 95% CI 0.99–1.03), as shown in Table II. Sensitivity analyses were performed in relation to outliers of the outcomes. The exclusion of implausible low values (sitting at age
