Psychological Medicine (2015), 45, 1779–1787. © Cambridge University Press 2014 doi:10.1017/S0033291714002906
OR I G I N A L A R T I C L E
Prenatal marijuana exposure, age of marijuana initiation, and the development of psychotic symptoms in young adults N. L. Day1*, L. Goldschmidt2, R. Day3, C. Larkby1 and G. A. Richardson1 1
Department of Psychiatry, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA University of Pittsburgh Medical Center, Pittsburgh, PA, USA 3 Department of Biostatistics, University of Pittsburgh Graduate School of Public Health, Pittsburgh, PA, USA 2
Background. Studies have demonstrated that an early age of onset of marijuana use (EAOM) is associated with a higher risk of developing psychotic symptoms (PS) compared to initiating marijuana use at a later age or not at all. Research has also found that prenatal marijuana exposure (PME) predicts EAOM. This report evaluates the relationships among PME, EAOM, and PS. Method. Subjects were initially interviewed in their fourth prenatal month. Women and offspring who completed the birth assessment (n = 763) were selected for follow-up. Women and their offspring were followed until the offspring were 22 years of age: 596 offspring were evaluated. At age 22, PS were assessed in the offspring with the Diagnostic Interview Schedule using DSM-IV criteria. Analyses controlled for significant covariates including other prenatal substance exposures, race, gender, and offspring substance use at 22 years. Results. PME and EAOM significantly predicted increased rates of PS at 22 years controlling for other significant covariates. The direct effect of PME on PS was marginally significant (p = 0.06) when EAOM was entered into the model and other covariates were fixed. In the mediation analysis, EAOM did not significantly mediate the association between PME and PS, controlling for significant covariates, nor was the indirect pathway significant when structural equation modeling was used. The total effect of the direct and indirect pathways was significant. Conclusions. In addition to EAOM, PME may also play a role in the association between marijuana use and the development of PS. This could highlight a new area for prevention. Received 24 July 2014; Revised 5 November 2014; Accepted 9 November 2014; First published online 23 December 2014 Key words: Early marijuana use, prenatal marijuana exposure, psychotic symptoms.
Introduction We have shown that prenatal marijuana exposure (PME) predicts early age of onset of marijuana use (EAOM) (Day et al. 2006), which is significantly associated with the development of psychotic symptoms (PS) and disorders (Arseneault et al. 2002; Zammit et al. 2002; Lynskey et al. 2003; Stefanis et al. 2004; Konings et al. 2008; Skinner et al. 2011). There is, however, no research on the association between PME and PS. This paper will explore whether there is an association between PME and PS and whether this association is mediated by EAOM. A recent national survey on prenatal drug use found that in 2004–2005, marijuana use was reported by 2.8% of pregnant women (SAMHSA, 2012). This rate,
* Address for correspondence: Dr N. L. Day, Department of Psychiatry, Webster Hall, Suite 108, 4415 Fifth Avenue, Pittsburgh, PA 15213, USA. (Email: [email protected])
however, is likely underestimated, and varies widely by age, race, and social class. The importance of studying the effects of PME is highlighted by several recent trends. One, recreational marijuana is legal in two states, medical marijuana is legal in 20 states (ONDCP, 2014) and both of these numbers are likely to increase. Two, marijuana use has increased steadily in the past decade while, at the same time, the belief that marijuana is harmful has decreased (SAMHSA, 2013). Three, the average potency of marijuana has increased from 3.4% Δ9-tetrahydrocannabinol (THC) in 1993 to 8.8% THC in 2008 (Mehmedic et al. 2010), and four, as marijuana becomes more available, the price will decrease, leading to an increase in use (Kilmer et al. 2010). Thus, it is likely that prenatal marijuana use will increase in the future and that the dose/ joint will be greater. Cannabinoid receptors appear in the fetal brain around the 14th week of gestation (Biegon & Kerman, 2001) and are concentrated in the hippocampus, cerebellum, amygdala, and cerebral cortex of the
1780 N. L. Day et al. brain (Fride, 2002), the areas that subserve cognitive and behavioral functioning. The cannabinoid receptors play an important role in the development of neuronal connectivity (Harkany et al. 2008) and intercellular signaling (Navarrete & Araque, 2008). Interference with endocannabinoid signaling during fetal development can lead to long-term changes in synaptic structure and function (Berghuis et al. 2007). Exogenous cannabinoids such as THC interact with the cannabinoid receptors and are known to affect brain development in animals (e.g. Kumar et al. 2001) and humans, leading to subsequent effects of PME including increased mood and anxiety symptoms (Gray et al. 2005), behavior, cognitive, and psychiatric problems (Day et al. 1994, 2011; Goldschmidt et al. 2000; Richardson et al. 2002; Larkby et al. 2007; Trezza et al. 2008). Early adolescence is also an important period in brain development. During this time, development occurs in the limbic structures of the brain and, in parallel, the number of cannabinoid CB1 receptors increases in the corticolimbic region (Spear, 2000; Giedd et al. 2006). In early adolescence, there is significant maturation in complex processes such as working memory (Luna et al. 2004), executive functioning, and emotional capacity (Yurgelun-Todd, 2007). Exogenous cannabinoids, such as THC, interfere with the functioning of the endogenous cannabinoid system (ECS) and may affect the maturation of the cortical neuronal networks during this vulnerable period (Bossong & Niesink, 2010), which may lead to a dysregulation of the ECS and a subsequent an increase in PS and psychosis (Malone et al. 2010). In the most recent U.S. Monitoring the Future study, rates of marijuana use were 16.5% in the 8th grade at age 13 years and 35.8% in the 10th grade at age 15 years (Johnston et al. 2014). Thus, a substantial number of adolescents initiate marijuana at an early age. There is a large body of work investigating the effects of marijuana use during early adolescence on the development of psychosis and PS. Analyses from the Dunedin study found that early marijuana users (by age 15) had more PS than did controls who were not early users, and a fourfold increase in schizophreniform disorders by age 26 (Arseneault et al. 2002). Studies have repeated this finding in a variety of populations (Zammit et al. 2002; Lynskey et al. 2003; Stefanis et al. 2004; Konings et al. 2008; Hides et al. 2009; Shapiro & Buckley-Hunter, 2010; Skinner et al. 2011). There was a dose–response association between level of marijuana use and the number of PS (Hides et al. 2009; Skinner, et al. 2011). Use of other drugs in this developmental period was not associated with PS (Zammit et al. 2002; Henquet et al. 2005). Earlier, we reported that PME predicted EAOM and frequency of marijuana use among 14-year-olds,
controlling for the offspring’s current alcohol and tobacco use, pubertal stage, sexual activity, delinquency, peer drug use, family history of drug abuse, and characteristics of the home environment, including parental depression, current drug use, and strictness/ supervision (Day et al. 2006). The current paper will explore the association between PME, EAOM, and PS. We will test three hypotheses: (1) PME will predict a higher rate of PS in exposed offspring at 22 years; (2) this association will remain significant after controlling for gender, race, prenatal exposure to other substances, and the young adult’s current use of alcohol, tobacco, and illicit drugs other than marijuana; and (3) this association will be mediated by EAOM.
Method Study design and sample selection In the Maternal Health Practices and Child Development Study (MHPCD), women were recruited from a prenatal clinic between 1982 and 1985, and interviewed at their fourth-month prenatal visit. This time point was selected to eliminate women with early fetal loss and elective abortions and women who came in late for prenatal care, and to ensure that we would have data on all of the study women at standardized times during pregnancy. Women were at least 18 years of age and English-speaking. All women attended the same general prenatal clinic and were not considered medically high-risk. A sequential sample of 1360 women was interviewed at this first phase. The refusal rate was 15%. Medical record review indicated that the women who refused did not differ from the participants on the basis of insurance status (private v. public), age, or race. In the initial sample of 1360 women, 30.1% used marijuana in the first trimester and 3.8% reported an average daily use (average joints/day) of 52 joints. From this initial sample, all women who used marijuana an average of 52 joints a month and a random sample of women who used less or not all were selected for the marijuana cohort. A parallel study of the effects of prenatal alcohol exposure (PAE) was also conducted using women from the initial cohort of 1360. In the PAE study, all women who drank 53 drinks per week and a random sample of all other women were selected for the sample. The two study cohorts had a 48% overlap and are combined for these analyses, resulting in a total cohort of 829 women. Women selected for the studies were interviewed again at the seventh prenatal month, birth, 8 and 18 months, 3, 6, 10, 14, 16, and 22 years. At delivery, 66 mother–child pairs were removed from the cohort, resulting in a birth cohort of 763
Marijuana use and the development of psychotic symptoms pairs. Losses included 18 fetal deaths, two twin births, one adoption, and 45 children who did not receive a physical examination at birth (21 delivered elsewhere, eight mothers refused the birth examination, and 16 were lost to follow-up). Twins were eliminated because size and physiological status at birth are affected in multiple births. At the 22-year follow-up, 608 young adult offspring, 80% of the birth cohort, remained in the study. Attrition was due to death (n = 11), adoption/institutionalization (n = 21), lost to follow-up (n = 56), refusal (n = 30), moved out of the Pittsburgh area (n = 29), and unable to participate due to low cognitive functioning (n = 8). There were no significant differences in prenatal marijuana, alcohol, or tobacco exposures or maternal socioeconomic status at birth between those included in the 22-year analyses (n = 608) and those who were not interviewed at this phase (n = 155). Eleven subjects did not complete the psychiatric interview, and one of the offspring reported no marijuana use, but tested positive for THC. These cases were excluded, resulting in a sample of 596 offspring for analysis.
Ethical standards Each phase of this research was approved by the Institutional Review Board of the University of Pittsburgh. A Certificate of Confidentiality provided assurance of confidentiality to the participants. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1974, as revised in 2008.
1781
measure of marijuana exposure, weighting them according to the amount of Δ9-THC in each (Gold, 1989; Julien, 1997). Offspring marijuana use at 22 years was assessed using the same instrument and summarized as average joints/day. Age of marijuana initiation At 14, 16, and 22 years, offspring were asked the age at which they first used marijuana. Use was defined as more than just a ‘puff’. Age of initiation reported at the 14-year phase was used in the analyses unless initiation had not yet occurred. For these latter subjects, age of initiation reported at the 16- or 22-year phase was used in the analyses. EAOM was defined as initiation prior to age 15. Thirty-eight percent of the offspring initiated use before age 15 and 83% had used marijuana by age 22. Other drug use: mothers and offspring In parallel with marijuana exposure, the usual, maximum, and minimum quantity and frequency were assessed for wine, beer, liquor, and wine and beer coolers. Alcohol use was summarized as average daily volume (ADV). We used the same method, described above for PME, to identify the three phases of first trimester use. Number, frequency, brand, and history of tobacco cigarettes were ascertained. Tobacco was expressed as the average number of cigarettes smoked per day. For other illicit drugs, the name and type of drug, quantity, frequency, pattern, and method of use were recorded but because it was infrequent, use was combined into a dichotomous variable (use/no use) for these analyses. At 22 years, the offspring were interviewed with the same instrument as the mothers.
Measures
Psychotic symptoms
Marijuana use
The PS variable was the sum of positive responses to 24 questions from the psychosis section of the Diagnostic Interview Schedule (DIS-IV; Robins et al. 2000). The symptoms included 14 possible delusional beliefs and 10 hallucinatory experiences. The assessors were trained to use the DIS and were blind to the subject’s prenatal exposure. Reliability of the interviewers was assessed frequently both by observation and by auditory recordings. The mean number of symptoms in the sample was 1.1 (S.D. = 1.8, range 0–15). For descriptive purposes, we categorized PS as none, 1 or 2, and 53.
Maternal marijuana use was measured for the year prior to pregnancy, across the first, second, and third trimesters, and for the previous year at subsequent follow-up phases. We developed a method to measure marijuana consumption from conception to pregnancy recognition, recognition to confirmation of pregnancy, and from confirmation to the end of the first trimester (Day et al. 1985). From this, we created measures of first trimester PME that are weighted by the amount of marijuana consumed during each of these three time periods and by the duration of each of these three time periods for each mother. We assessed the minimum, usual, and maximum quantity and frequency of use. We asked about marijuana, sinsemilla, and hashish and combined these reports in our
Analysis Negative binomial regression models were applied to assess the effects of PME on the rate of PS, controlling
1782 N. L. Day et al. for significant covariates. In the Poisson distribution, the expected mean and variance are assumed to be equal and the negative binomial was applied to adjust for over-dispersion. The regression coefficients and dispersion parameters were estimated using maximum likelihood. To assess overall goodness-of-fit, the deviance parameter was compared to a χ2 distribution with degrees of freedom equal to the model residual. Covariates included prenatal exposure to alcohol and tobacco, gender, race, and current use of alcohol, tobacco, and illicit drugs. The analyses were conducted hierarchically to assess the effect of EAOM on the association between PME and PS. Models were run excluding and including EAOM. Statistical tests of mediation are not available within the negative binomial regression framework. Therefore, in addition to the above analyses, we dichotomized PS to 53 symptoms and tested whether the indirect effects of PME on PS via EAOM was significant using structural equation modeling (SEM). A logit regression was used to estimate the pathways and mediation was tested as the product of the pathways from PME to EAOM and from EAOM to PS.
had at least one child. At 22 years, 50% of the offspring used marijuana currently, 92% drank alcohol, 43% smoked tobacco cigarettes, and 17% used illicit drugs other than marijuana. Characteristics of prenatal marijuana users Women who used marijuana during the first trimester were more likely to be single, to have a lower family income, and to use more alcohol and other illicit drugs than women who did not use marijuana in the first trimester (Table 1). These two groups did not differ in age, education, or first trimester tobacco use. Characteristics of offspring with PME Offspring with first trimester marijuana exposure were more likely to be African American, younger, and to use more tobacco at 22 years of age than the nonexposed offspring. The two groups of offspring did not differ in gender, education, work or marital status, or current alcohol, marijuana, or other illicit drug use (Table 1). Does PME predict a higher rate of PS in exposed offspring at 22 years?
Results Study sample description In the first trimester, 74% of the women had completed high school, 61% earned less than $400 per month, the median age was 22 years (range 18–42), 33% of the women were primigravidous, and 68% were not married. At the first trimester, 41% of the women reported using marijuana, 64% drank alcohol, and 53% smoked cigarettes. The mean first trimester marijuana use among users was 0.9 joints/day (range 0.001–7), mean alcohol use was 0.9 drinks/day (range 0.006–20), and mean number of cigarettes was 15.3/day (range 0.5–50). Twelve percent of the women reported illicit drug use other than marijuana, including 4.0% who reported cocaine use. Average birth weight of the offspring was 3198 g (7 lb) (range 1040–4990 g), 8% were premature (
OR I G I N A L A R T I C L E
Prenatal marijuana exposure, age of marijuana initiation, and the development of psychotic symptoms in young adults N. L. Day1*, L. Goldschmidt2, R. Day3, C. Larkby1 and G. A. Richardson1 1
Department of Psychiatry, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA University of Pittsburgh Medical Center, Pittsburgh, PA, USA 3 Department of Biostatistics, University of Pittsburgh Graduate School of Public Health, Pittsburgh, PA, USA 2
Background. Studies have demonstrated that an early age of onset of marijuana use (EAOM) is associated with a higher risk of developing psychotic symptoms (PS) compared to initiating marijuana use at a later age or not at all. Research has also found that prenatal marijuana exposure (PME) predicts EAOM. This report evaluates the relationships among PME, EAOM, and PS. Method. Subjects were initially interviewed in their fourth prenatal month. Women and offspring who completed the birth assessment (n = 763) were selected for follow-up. Women and their offspring were followed until the offspring were 22 years of age: 596 offspring were evaluated. At age 22, PS were assessed in the offspring with the Diagnostic Interview Schedule using DSM-IV criteria. Analyses controlled for significant covariates including other prenatal substance exposures, race, gender, and offspring substance use at 22 years. Results. PME and EAOM significantly predicted increased rates of PS at 22 years controlling for other significant covariates. The direct effect of PME on PS was marginally significant (p = 0.06) when EAOM was entered into the model and other covariates were fixed. In the mediation analysis, EAOM did not significantly mediate the association between PME and PS, controlling for significant covariates, nor was the indirect pathway significant when structural equation modeling was used. The total effect of the direct and indirect pathways was significant. Conclusions. In addition to EAOM, PME may also play a role in the association between marijuana use and the development of PS. This could highlight a new area for prevention. Received 24 July 2014; Revised 5 November 2014; Accepted 9 November 2014; First published online 23 December 2014 Key words: Early marijuana use, prenatal marijuana exposure, psychotic symptoms.
Introduction We have shown that prenatal marijuana exposure (PME) predicts early age of onset of marijuana use (EAOM) (Day et al. 2006), which is significantly associated with the development of psychotic symptoms (PS) and disorders (Arseneault et al. 2002; Zammit et al. 2002; Lynskey et al. 2003; Stefanis et al. 2004; Konings et al. 2008; Skinner et al. 2011). There is, however, no research on the association between PME and PS. This paper will explore whether there is an association between PME and PS and whether this association is mediated by EAOM. A recent national survey on prenatal drug use found that in 2004–2005, marijuana use was reported by 2.8% of pregnant women (SAMHSA, 2012). This rate,
* Address for correspondence: Dr N. L. Day, Department of Psychiatry, Webster Hall, Suite 108, 4415 Fifth Avenue, Pittsburgh, PA 15213, USA. (Email: [email protected])
however, is likely underestimated, and varies widely by age, race, and social class. The importance of studying the effects of PME is highlighted by several recent trends. One, recreational marijuana is legal in two states, medical marijuana is legal in 20 states (ONDCP, 2014) and both of these numbers are likely to increase. Two, marijuana use has increased steadily in the past decade while, at the same time, the belief that marijuana is harmful has decreased (SAMHSA, 2013). Three, the average potency of marijuana has increased from 3.4% Δ9-tetrahydrocannabinol (THC) in 1993 to 8.8% THC in 2008 (Mehmedic et al. 2010), and four, as marijuana becomes more available, the price will decrease, leading to an increase in use (Kilmer et al. 2010). Thus, it is likely that prenatal marijuana use will increase in the future and that the dose/ joint will be greater. Cannabinoid receptors appear in the fetal brain around the 14th week of gestation (Biegon & Kerman, 2001) and are concentrated in the hippocampus, cerebellum, amygdala, and cerebral cortex of the
1780 N. L. Day et al. brain (Fride, 2002), the areas that subserve cognitive and behavioral functioning. The cannabinoid receptors play an important role in the development of neuronal connectivity (Harkany et al. 2008) and intercellular signaling (Navarrete & Araque, 2008). Interference with endocannabinoid signaling during fetal development can lead to long-term changes in synaptic structure and function (Berghuis et al. 2007). Exogenous cannabinoids such as THC interact with the cannabinoid receptors and are known to affect brain development in animals (e.g. Kumar et al. 2001) and humans, leading to subsequent effects of PME including increased mood and anxiety symptoms (Gray et al. 2005), behavior, cognitive, and psychiatric problems (Day et al. 1994, 2011; Goldschmidt et al. 2000; Richardson et al. 2002; Larkby et al. 2007; Trezza et al. 2008). Early adolescence is also an important period in brain development. During this time, development occurs in the limbic structures of the brain and, in parallel, the number of cannabinoid CB1 receptors increases in the corticolimbic region (Spear, 2000; Giedd et al. 2006). In early adolescence, there is significant maturation in complex processes such as working memory (Luna et al. 2004), executive functioning, and emotional capacity (Yurgelun-Todd, 2007). Exogenous cannabinoids, such as THC, interfere with the functioning of the endogenous cannabinoid system (ECS) and may affect the maturation of the cortical neuronal networks during this vulnerable period (Bossong & Niesink, 2010), which may lead to a dysregulation of the ECS and a subsequent an increase in PS and psychosis (Malone et al. 2010). In the most recent U.S. Monitoring the Future study, rates of marijuana use were 16.5% in the 8th grade at age 13 years and 35.8% in the 10th grade at age 15 years (Johnston et al. 2014). Thus, a substantial number of adolescents initiate marijuana at an early age. There is a large body of work investigating the effects of marijuana use during early adolescence on the development of psychosis and PS. Analyses from the Dunedin study found that early marijuana users (by age 15) had more PS than did controls who were not early users, and a fourfold increase in schizophreniform disorders by age 26 (Arseneault et al. 2002). Studies have repeated this finding in a variety of populations (Zammit et al. 2002; Lynskey et al. 2003; Stefanis et al. 2004; Konings et al. 2008; Hides et al. 2009; Shapiro & Buckley-Hunter, 2010; Skinner et al. 2011). There was a dose–response association between level of marijuana use and the number of PS (Hides et al. 2009; Skinner, et al. 2011). Use of other drugs in this developmental period was not associated with PS (Zammit et al. 2002; Henquet et al. 2005). Earlier, we reported that PME predicted EAOM and frequency of marijuana use among 14-year-olds,
controlling for the offspring’s current alcohol and tobacco use, pubertal stage, sexual activity, delinquency, peer drug use, family history of drug abuse, and characteristics of the home environment, including parental depression, current drug use, and strictness/ supervision (Day et al. 2006). The current paper will explore the association between PME, EAOM, and PS. We will test three hypotheses: (1) PME will predict a higher rate of PS in exposed offspring at 22 years; (2) this association will remain significant after controlling for gender, race, prenatal exposure to other substances, and the young adult’s current use of alcohol, tobacco, and illicit drugs other than marijuana; and (3) this association will be mediated by EAOM.
Method Study design and sample selection In the Maternal Health Practices and Child Development Study (MHPCD), women were recruited from a prenatal clinic between 1982 and 1985, and interviewed at their fourth-month prenatal visit. This time point was selected to eliminate women with early fetal loss and elective abortions and women who came in late for prenatal care, and to ensure that we would have data on all of the study women at standardized times during pregnancy. Women were at least 18 years of age and English-speaking. All women attended the same general prenatal clinic and were not considered medically high-risk. A sequential sample of 1360 women was interviewed at this first phase. The refusal rate was 15%. Medical record review indicated that the women who refused did not differ from the participants on the basis of insurance status (private v. public), age, or race. In the initial sample of 1360 women, 30.1% used marijuana in the first trimester and 3.8% reported an average daily use (average joints/day) of 52 joints. From this initial sample, all women who used marijuana an average of 52 joints a month and a random sample of women who used less or not all were selected for the marijuana cohort. A parallel study of the effects of prenatal alcohol exposure (PAE) was also conducted using women from the initial cohort of 1360. In the PAE study, all women who drank 53 drinks per week and a random sample of all other women were selected for the sample. The two study cohorts had a 48% overlap and are combined for these analyses, resulting in a total cohort of 829 women. Women selected for the studies were interviewed again at the seventh prenatal month, birth, 8 and 18 months, 3, 6, 10, 14, 16, and 22 years. At delivery, 66 mother–child pairs were removed from the cohort, resulting in a birth cohort of 763
Marijuana use and the development of psychotic symptoms pairs. Losses included 18 fetal deaths, two twin births, one adoption, and 45 children who did not receive a physical examination at birth (21 delivered elsewhere, eight mothers refused the birth examination, and 16 were lost to follow-up). Twins were eliminated because size and physiological status at birth are affected in multiple births. At the 22-year follow-up, 608 young adult offspring, 80% of the birth cohort, remained in the study. Attrition was due to death (n = 11), adoption/institutionalization (n = 21), lost to follow-up (n = 56), refusal (n = 30), moved out of the Pittsburgh area (n = 29), and unable to participate due to low cognitive functioning (n = 8). There were no significant differences in prenatal marijuana, alcohol, or tobacco exposures or maternal socioeconomic status at birth between those included in the 22-year analyses (n = 608) and those who were not interviewed at this phase (n = 155). Eleven subjects did not complete the psychiatric interview, and one of the offspring reported no marijuana use, but tested positive for THC. These cases were excluded, resulting in a sample of 596 offspring for analysis.
Ethical standards Each phase of this research was approved by the Institutional Review Board of the University of Pittsburgh. A Certificate of Confidentiality provided assurance of confidentiality to the participants. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1974, as revised in 2008.
1781
measure of marijuana exposure, weighting them according to the amount of Δ9-THC in each (Gold, 1989; Julien, 1997). Offspring marijuana use at 22 years was assessed using the same instrument and summarized as average joints/day. Age of marijuana initiation At 14, 16, and 22 years, offspring were asked the age at which they first used marijuana. Use was defined as more than just a ‘puff’. Age of initiation reported at the 14-year phase was used in the analyses unless initiation had not yet occurred. For these latter subjects, age of initiation reported at the 16- or 22-year phase was used in the analyses. EAOM was defined as initiation prior to age 15. Thirty-eight percent of the offspring initiated use before age 15 and 83% had used marijuana by age 22. Other drug use: mothers and offspring In parallel with marijuana exposure, the usual, maximum, and minimum quantity and frequency were assessed for wine, beer, liquor, and wine and beer coolers. Alcohol use was summarized as average daily volume (ADV). We used the same method, described above for PME, to identify the three phases of first trimester use. Number, frequency, brand, and history of tobacco cigarettes were ascertained. Tobacco was expressed as the average number of cigarettes smoked per day. For other illicit drugs, the name and type of drug, quantity, frequency, pattern, and method of use were recorded but because it was infrequent, use was combined into a dichotomous variable (use/no use) for these analyses. At 22 years, the offspring were interviewed with the same instrument as the mothers.
Measures
Psychotic symptoms
Marijuana use
The PS variable was the sum of positive responses to 24 questions from the psychosis section of the Diagnostic Interview Schedule (DIS-IV; Robins et al. 2000). The symptoms included 14 possible delusional beliefs and 10 hallucinatory experiences. The assessors were trained to use the DIS and were blind to the subject’s prenatal exposure. Reliability of the interviewers was assessed frequently both by observation and by auditory recordings. The mean number of symptoms in the sample was 1.1 (S.D. = 1.8, range 0–15). For descriptive purposes, we categorized PS as none, 1 or 2, and 53.
Maternal marijuana use was measured for the year prior to pregnancy, across the first, second, and third trimesters, and for the previous year at subsequent follow-up phases. We developed a method to measure marijuana consumption from conception to pregnancy recognition, recognition to confirmation of pregnancy, and from confirmation to the end of the first trimester (Day et al. 1985). From this, we created measures of first trimester PME that are weighted by the amount of marijuana consumed during each of these three time periods and by the duration of each of these three time periods for each mother. We assessed the minimum, usual, and maximum quantity and frequency of use. We asked about marijuana, sinsemilla, and hashish and combined these reports in our
Analysis Negative binomial regression models were applied to assess the effects of PME on the rate of PS, controlling
1782 N. L. Day et al. for significant covariates. In the Poisson distribution, the expected mean and variance are assumed to be equal and the negative binomial was applied to adjust for over-dispersion. The regression coefficients and dispersion parameters were estimated using maximum likelihood. To assess overall goodness-of-fit, the deviance parameter was compared to a χ2 distribution with degrees of freedom equal to the model residual. Covariates included prenatal exposure to alcohol and tobacco, gender, race, and current use of alcohol, tobacco, and illicit drugs. The analyses were conducted hierarchically to assess the effect of EAOM on the association between PME and PS. Models were run excluding and including EAOM. Statistical tests of mediation are not available within the negative binomial regression framework. Therefore, in addition to the above analyses, we dichotomized PS to 53 symptoms and tested whether the indirect effects of PME on PS via EAOM was significant using structural equation modeling (SEM). A logit regression was used to estimate the pathways and mediation was tested as the product of the pathways from PME to EAOM and from EAOM to PS.
had at least one child. At 22 years, 50% of the offspring used marijuana currently, 92% drank alcohol, 43% smoked tobacco cigarettes, and 17% used illicit drugs other than marijuana. Characteristics of prenatal marijuana users Women who used marijuana during the first trimester were more likely to be single, to have a lower family income, and to use more alcohol and other illicit drugs than women who did not use marijuana in the first trimester (Table 1). These two groups did not differ in age, education, or first trimester tobacco use. Characteristics of offspring with PME Offspring with first trimester marijuana exposure were more likely to be African American, younger, and to use more tobacco at 22 years of age than the nonexposed offspring. The two groups of offspring did not differ in gender, education, work or marital status, or current alcohol, marijuana, or other illicit drug use (Table 1). Does PME predict a higher rate of PS in exposed offspring at 22 years?
Results Study sample description In the first trimester, 74% of the women had completed high school, 61% earned less than $400 per month, the median age was 22 years (range 18–42), 33% of the women were primigravidous, and 68% were not married. At the first trimester, 41% of the women reported using marijuana, 64% drank alcohol, and 53% smoked cigarettes. The mean first trimester marijuana use among users was 0.9 joints/day (range 0.001–7), mean alcohol use was 0.9 drinks/day (range 0.006–20), and mean number of cigarettes was 15.3/day (range 0.5–50). Twelve percent of the women reported illicit drug use other than marijuana, including 4.0% who reported cocaine use. Average birth weight of the offspring was 3198 g (7 lb) (range 1040–4990 g), 8% were premature (
