Prenatal Methamphetamine Exposure: Effects on Child Development

Abstract

Background

In Germany, the 12-month prevalence of methamphetamine use among persons aged 15 to 34 is 1.9%. An increasing number of newborns are being born after a prenatal methamphetamine exposure (PME). In 2014, in the German state of Saxony, approximately four out of 1000 newborns were affected.

Methods

This systematic review (Prospero registration number CRD42017060536) includes publications that were published between January 1990 and November 2019. The purpose was to determine the effects of PME on the peri- and neonatal condition of the affected children and on their further long-term development. Observational studies with a control group were included in the review and examined for their methodological quality.

Results

31 publications, which dealt with two prospective and six retrospective cohort studies, were included in the review. The studies involved a total of 4446 mother–child pairs with PME, compared with 43 778 pairs without PME. A meta-analysis revealed that PME was associated with, among other findings, lower birth weight (SMD = –0.348; 95% confidence interval [-0.777; 0.081]), shorter body length (SMD= –0.198 [-0.348; -0.047]), and smaller head circumference (SMD= –0.479 [-1.047; 0.089]). Some differences between the groups with and without PME persist into the toddler years. Moreover, children with PME much more commonly display psychological and neurocognitive abnormalities, which are more severe in children growing up in problematic surroundings (discord, violence, poverty, low educational level of the parent or caregiver). A limitation of this review is that not all studies employed an objective or quantitative measure of methamphetamine use.

Conclusion

The documented effects of PME on child development necessitate early treatment of the affected expectant mothers, children, and families. Emphasis should be placed on structured and interdisciplinary preventive measures for methamphetamine use.

Compared to the European Union (EU), Germany (D) holds a middle-ranking position with regard to the lifetime prevalence of methamphetamine (MA) use in the 15 to 64 years age group (D: 3.6%, EU: 3.7%); in the twelve-month prevalence among the 15– to 34-year-olds, Germany occupies a top position (D: 1.9%, EU: 1.0%) (1). Given its high addictive potential and its popularity especially among young adults, MA is gaining increasing attention among the substances regulated by the German Narcotic Drugs Act (Betäubungsmittelgesetz, BtMG) (2). Consequently, the 2019 report of the Drug Commissioner of the Federal Government of Germany highlights above all the need for prevention (2).

MA use is associated with high-risk sexual behavior and early pregnancies (3). Reliable data on the prevalence of amphetamine use during pregnancy are scarce. In an English study (N = 149), urine of pregnant women was tested for substances, including amphetamines; positive amphetamine results were found in n = 5, corresponding to an absolute share of 3.3% and about one-third (31.3%) of the women who tested positive for drugs (4). In the state of Saxony, the number of hospitalized newborns with prenatal MA exposure (PME) significantly increased over a period of four years from below 1/1000 newborns (2010) to about 4/1000 newborns (2014) (5).

In recent years, children with PME have been followed up systematically. These studies focused on the effects of PME on the physical and neurocognitive development in early childhood. Some of these studies were considered in the clinical practice (S3) guideline “Methamphetamine-related Disorders“ (6). The aim of this review is to systematically summarize the available data on the effects of PME on child development and to conduct a meta-analysis where appropriate.

Methods

This systematic review addresses the following question (7): What are the effects of prenatal MA exposure on the peri- and neonatal condition and long-term development of the affected children?

The inclusion criteria were defined based on this research question and are listed in etable 1. Neuroimaging studies were not included.

eTable 1. Inclusion criteria.

Category	Inclusion criteria
Population	Pregnant women or mother-child pairs
Exposure	Methamphetamine use in pregnancy
Outcomes	Somatic, psychological, psychosocial, and developmental outcomes which were measured in the unborn/born child or collected in a standardized way
Design	Observational studies: Case series (n ≥ 6), cross-sectional studies, case-control studies, and cohort studies with control group
Publication type	Original articles in German or English published in scientific journals, with online abstract

The review protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) database (ID: CRD42017060536). Detailed information about the search method and data extraction is presented in the eMethods 1 section, the search strings in the eMethods 2 section.

Study quality was assessed using the Critical Appraisal Skills Program (CASP) tool (8). Two reviewers conducted the methodological quality appraisal independently; in the absence of a consensus, a third reviewer was consulted.

The review results were reported in accordance with the „Preferred Reporting Items for Systematic Reviews and Meta-Analyses“ (PRISMA) statement (9).

Results

Study overview

In total, 681 publications were identified through database search and additional ten through manual search. After exclusion of duplicates and title abstract screening, 38 publications were included in the full-text screening. After exclusion of seven further publications, 31 publications were included in this systematic review (figure 1). A list of excluded publications, together with the reasons for exclusion, can be found in the eMethods 1 section.

Figure 1.

Flowchart of the systematic review

Descriptive information about the studies

The study by van Dyk et al. (10) as well as the Infant Development, Environment, and Lifestyle (IDEAL) study (which is covered in the publications [11–34]) are prospective cohort studies. Furthermore, new subpopulations were sampled from the IDEAL study population to address specific research questions. Consequently, overlapping may occur between mother-infant pairs. For each study time point and outcome, only the respective first IDEAL publication was used as the basis. This results in a study duration of up to seven and a half years (16, 32). Across all 25 cohort (sub-) studies, 3729 mother-infant pairs with PME were compared to 8717 pairs without PME (etable 2).

eTable 2. Descriptive information about the studies included in this review.

Author, year, country	Study type	Study duration	Follow-up(s) at	Exposure measurement	n exposed	n non-exposed	Matching variables and confounders	Study quality using CASP tool
Infant Development. Environment. and Lifestyle (IDEAL) study
Abar et al. 2013, USA (11)	Prospective cohort study	6.5 years	5 years 6.5 years	Meconium analysis and maternal self-report	162	185	Maternal Ethnicity (USA) Schooling < high school Insurance status (private versus public) Newborn Birth weight Confounders Prenatal consumption of alcohol, tobacco and marihuana	21 out of 24 possible points
Abar et al. 2014, USA/ New Zealand (NZ) (12)		3 years	1 year 2 years 3 years		USA: 204NZ: 108	USA: 208NZ: 115		21 out of 24 possible points
Derauf et al. 2011, USA (13)		3 years	1 month 1 year 2 years 2.5 years 3 years		142	148		23 out of 24 possible points
Derauf et al. 2012, USA (14)		5 years	1 month 1 year 2 years 2.5 years 3 years 5 years 5.5 years		20	15		15 out of 24 possible points
Derauf et al. 2012, USA (15)		5.5 years	1 month 1 year 2 years 2.5 years 3 years 5 years 5.5 years		133	130		21 out of 24 possible points
Diaz et al. 2014, USA (16)		7.5 years	7.5 years		151	147		23 out of 24 possible points
Eze et al. 2016, USA (32)		7.5 years	7.5 years		146	144		20 out of 24 possible points
Galland et al. 2013, USA (17)		3 months	3 months		42	57		19 out of 24 possible points
Himes et al. 2014, USA (18)		6.5 years	6.5 years		131	133		19 out of 24 possible points
Kiblawi et al. 2013, USA (19)		5.5 years	5.5 years		153	148		21 out of 24 possible points
Kiblawi et al. 2014, USA (20)		5 days	5 days		185	195		18 out of 24 possible points
La Gasse et al. 2011, USA/New Zealand (21)		5 days	5 days		USA: 183 NZ: 85	USA: 196 NZ: 95		20 out of 24 possible points
La Gasse et al. 2012, USA (22)	Prospective cohort study	5 years	5 years	Meconium analysis and maternal self-report	166	164	Maternal Ethnicity (USA) Schooling < high school Insurance status (private versus public) Newborn Birth weight Confounders Prenatal consumption of alcohol, tobacco and marihuana	22 out of 24 possible points
Liles et al. 2012, USA (23)		3 years	1 month 1 year 2 years 2.5 years 3 years		75	137		22 out of 24 possible points
Nguyen et al. 2010, USA (24)		Not stated	Not stated		204	3 501		18 out of 24 possible points
Paz et al. 2009, USA (25)		5 days	5 days		50	86		20 out of 24 possible points
Shah et al. 2012, USA (33)		Not stated	Not stated		204	208		21 out of 24 possible points
Smith et al. 2006, USA (26)		Not stated	Not stated		84	1 534		19 out of 24 possible points
Smith et al. 2008, USA (27)		5 days	5 days		74	92		21 out of 24 possible points
Smith et al. 2011, USA (28)		3 years	1 month 1 year 2 years 2.5 years 3 years		179	177		20 out of 24 possible points
Smith et al. 2012, USA (29)		1 month	1 month		126	193		20 out of 24 possible points
Twomey et al. 2013, USA (30)		5 years	5 years		214	198		23 out of 24 possible points
Wouldes et al. 2014, USA (34)		3 years	1 year 2 years 3 years		103	107		20 out of 24 possible points
Zabaneh et al. 2012, USA (31)		3 years	1 year 2 years 3 years		186	175	Maternal Ethnicity (USA) Schooling < high school Insurance status (private versus public) Newborn Birth weight Confounders Prematurity, place of recruitment, relationship status	20 out of 24 possible points
Other studies, not related to IDEAL
Chomchai et al. 2004, Thailand (35)	Retrospective cohort study	Not stated	Not stated	Maternal self-report	47	49	Maternal Age Newborn Sex Birth weight Size at birth Head circumference at birth 1-minute Apgar score 5-minute Apgar score Gestational age Confounders Prenatal consumption of alcohol, tobacco and marihuana	17 out of 24 possible points
van Dyk et al. 2014, South Africa (10)	Prospective longitudinal cohort study	4 years	2 years 4 years	Maternal self-report	15	21	Maternal Mother working Socioeconomic status Newborn Age Sex Ethnicity (South Africa) Confounders Not stated	15 out of 24 possible points
Good et al. 2010, USA (36)	Retrospective cohort study	Not relevant	Not relevant	Urine test Maternal self-report	273	34 055	Maternal Age < 20 Number of prenatal checks < 5 Ethnicity hispanic (usa) Relationship status married Confounders not stated	17 out of 24 possible points
Pflügner et al. 2018, Germany (40)		Not relevant	Not relevant	Drug test Maternal self-report	102	171	Newborn Date of birth Sex Confounders Year	22 out of 24 possible points
Saleh Gargari et al. 2012, Iran (38)		Not relevant	Not relevant	Maternal self-report	17	519	Maternal Age First pregnancy Number of prenatal checks Anemia Caesarean delivery Confounders Not stated	13 out of 24 possible points
Smith et al. 2003, USA (37)	Retrospective cohort study	Not relevant	Not relevant	Urine test Maternal self-report	134	160	Maternal Age Complaints during pregnancy Abortions Ethnicity (USA) Number of prenatal checks < 5 Prenatal tobacco consumption Prenatal marihuana consumption Confounders Prenatal alcohol consumption	15 out of 24 possible points
Wright et al. 2015, USA (39)		Not relevant	Not relevant	Random toxicological testing Maternal self-report	144	107	Not stated Confounders Number of prenatal checks < 5 Chronic hypertension Late gestosis Diabetes Trimenon of last crystal consumption Prenatal consumption of any other illegal drug except crystal	15 out of 24 possible points

CASP, Critical Appraisal Skills Program (8)

The remaining six studies are retrospective cohort studies based on patient records of newborns and their mothers (35– 40). Here, 717 mother-infant pairs with PME were compared to 35 061 pairs without PME (etable 2).

Most studies investigate children in the United States (11, 13– 17, 19, 22, 25– 28, 30– 33, 36, 37, 39); three IDEAL studies look at cohorts in New Zealand (12, 21, 34). In addition, studies from Thailand (35), South Africa (10), Iran (38), and Germany (40) were included.

With the exception of the studies by Zabaneh et al. (31), van Dyk et al. (10), Good et al. (36), Pflügner et al. (40), and Saleh-Gargari et al. (38), all studies adjusted for the use of other addictive substances, including, in particular, alcohol, tobacco and marihuana (eTable 2, eBox 1).

eBOX 1. Methodological details of the individual studies.

• Matching of exposure group and control group

In all IDEAL studies, matching of exposure group and control group was performed based on the ethnicity of the mother, her educational attainment (above or below high school level) and the insurance status (private or public, i.e. Medicaid in den United States) as well as for the child’s birth weight (11– 34). In the remaining studies, further characteristics of the newborn (such as Apgar Score or sex) (10, 35) were also included as well as detailed information about the mother, such as her other drug use or the course of the pregnancy (36– 38). In one study, no matching was performed (39).

• Confounding variables

Five studies did not provide information about potential confounding variables which could have had an effect on the outcome (so-called covariates), in addition to matching variables (10, 17, 35– 38). A standardized covariate set was developed for the IDEAL studies over the course of the study period. In addition to the above-mentioned matching variables, maternal age, her relationship status and socioeconomic status, the utilization of screening examinations were also included, as well as the use of other drugs pregnancy (alcohol, tobacco and marihuana, in some cases also cocaine [21, 27, 29]) (11– 16, 18– 33). In later follow-up studies, further neonatal characteristics of the child were added, such as gestational age and head circumference (for example 11) as well as in some studies characteristics of the guardians (not the natural parents) which may have an effect in particular on the cognitive or behavioral development of the child, including stress during parenting or physical abuse by the partner (for example [30]). In some studies, confounding variables in the child’s environment, such as quality of the residential area (for example [19]), are also covered..

• Matching variables (eTable 5)

Effects of PME, both cognitive and physical, were noted in the various phases of child development (figure 2).

Figure 2.

Effects of prenatal methamphetamine exposure were found in the various phases of child development. PME, prenatal methamphetamine exposure

Effects of PME on postnatal condition

Pregnant women with MA use had their first prenatal visit five weeks later than pregnant women without MA use (26). Infants with PME were less mature at birth; in five of six studies, the gestational age of newborns with PME was significantly lower compared to newborns without PME; the range of age differences was 0.3 to 2.2 weeks (26, 37– 40). The effect of PME on gestational age, as revealed in the meta-analysis, is shown in eFigure 1 based on the standardized mean difference (SMD; SMD = -0.613; 95% confidence interval: [-1.448; 0.222]).

eFigure 1.

Newborns with PME had a statistically significantly lower gestational age compare to newborns without PME.

SMD, standardized mean difference; 95% CI, 95% confidence interval

Newborns with PME were in a poorer postnatal condition; their Apgar scores were lower after one minute in three of three studies (25, 35, 37) (SMD = –0.166 [-0.458; 0.125]) and after five minutes in two of four studies (25, 35, 37, 38). In the two other studies, the 5-minute Apgar score was equal or higher (35, 38) (SMD = 0.029 [0.470; 0.528]) (eFigures 2, 3). Newborns and infants with PME received inpatient treatment more commonly (odds ratio = 2.66 [2.339; 2.991]; 35-times in the exposed group [n = 204], 15-times in the control group [n = 208]) (33), and over a longer period of time (38, 40). One study found an increase in mortality among newborns with PME. In the exposed group, the mortality rate (with an observation period based on six years of registry data) was about 5% (n = 11 of 237), in the control group about 1% (n = 325 of 34 055) (36).

In the IDEAL study, newborn behavior was assessed by blinded examiners, using the Neonatal Intensive Care Unit Network Neurobehavioral Scale (NNNS) at postnatal day 5 and at one month postpartum, (27, 29). Newborns with PME showed more physiological signs of stress (g = 0.212 [-0.095; 0.519]); at the same time they were more likely to be lethargic (g = 0.167 [-0.159; 0.493]) and more difficult to arouse (g = -0.327 [-0.654; 0]) (27). At one month postpartum, newborns with PME had a decreased ability to self-regulate (g = -1.096 [-1.338; 0.853]), were still more difficult to arouse (g = 2.473 [2.176; 2.771] and needed more support during the examination (g = 0.455 [-0.221; 0.689]) (29). The association was confirmed after adjusting for socioeconomic status (27, 29). Frequent MA use of the pregnant women (>3 times a week) was associated with lower excitability scores and increased lethargy in the newborn (21). PME had no significant effect on reflexes, quality of movement and attention– neither immediately nor at one month postpartum (27, 29).

The effects of PME on intrauterine growth were noticed at birth. Neonates with PME were found to have lower values than the control group in six studies for:

• Birth weight (SMD = -0.348 [-0.777; 0.081]) (27, 34, 35, 37– 39)

• Length (SMD = -0.198 [-0.348; -0.047]) (27, 34, 35, 37– 39) and

• Head circumference (SMD = -0.479 [-1.047; 0.089] (27, 34, 35, 37– 39) (figure 3).

Figure 3.

At birth, effects of prenatal methamphetamine exposure on intrauterine growth were noted

SMD, standardized mean difference; 95% CI, 95% confidence interval

Pflügner et al. found similar effects in a German sample based on statistically significant differences in the (for gestational age and sex adapted) standard deviation scores between the exposed group and the control group (40). In addition, Pflügner et al. found respiratory disorders, such as disorders of adaptation (31/102 of the exposed infants versus 8/171 of the non-exposed infants; relative risk [RR] = 6.496 [6.12; 6.972]) and flaccid muscle tone (7/102 of the exposed infants versus none of the 171 non-exposed infants) (40).

Children with PME had a lower body weight during the first years of life; however, these difference were statistically significant at the age of two years (g = -0.3 [-0.59; -0.02]), but not at the age of one and three years (one year: g = -0.14 [-0.43; 0.14]; three years: g = -0.16 [-0.46; 0.14]) (34). The height at the age of three was statistically significantly shorter in children with PME (g = -0.40 [-0.71; -0.1]), but not at the age of one or two years (one year: g = -0.2 [-0.49; 0.08], two years: g = -0.2 [-0.48; 0.08]) (34). Head circumference at the age of one year was still statistically significantly smaller in children with PME (g = -0.3 [-0.58; -0.01]), but then converged to the control group (two years: g = -0.14 [-0.42; 0.14], three years: g = -0.12 [-0.42; 0.18]) (34).

Effects of PME on neurocognitive and motor development

At various points in time, evidence of negative effects of PME on cognitive development was found. At the age of 4 years, children with PME performed poorer on social ability and hand and eye coordination and had a lower overall performance on the Griffiths Mental Development Scales; in contrast, these children had no significantly greater difficulties with locomotion, practical reasoning as well as hearing and speech (10). Furthermore, at the age of 7.5 years, children with PME had greater cognitive deficits, such as learning and concentration difficulties (16). These results persisted even after adjusting for socioeconomic status and maternal educational attainment (10, 14, 16).

Effects on attention were observed at the age of 5.5 years. Children with PME had a higher reaction time in the Connors‘ Continuous Performance (CCP) test for the diagnosis of attention-deficit/hyperactivity disorder (ADHD) (14) and were less accurate in their reactions in the Dots test (15). This also held true after adjustment for socioeconomic status and maternal education (15). The reaction time according to the CCP test was statically significantly higher for children with frequent prenatal ME exposure (>3 times per week) (g = 0.097 [-0.331; 0.525]). In addition, they made more mistakes in the test (g = -0.5551 [-0.984; -0.119]) (15).

At the age of one year, children with PME performed poorer in grasping tasks (28); one study showed lower scores in the Psychomotor Development Index—an effect that remained stable until two years of age (34). Using the Mental Development Index, however, no difference between children aged 1 to 3 years with PME and without PME was found (28, 34).

Other effects of PME

Later in their lives, children with PME were more likely to have contact with youth protection agencies (109/204 of the exposed versus 5/208 of the non-exposed children; RR = 22.228 [21.781; 22.675]) (33). In addition, a chronic increase in blood pressure was significantly more common among children with PME (11/144 of the exposed versus 2/107 of the non-exposed children; RR = 4.087 [3.329; 4.845]) (39). In the sample studied by Pflügner et al., 20 of the 107 exposed children showed signs of neonatal abstinence syndrome (34).

Galland et al. found no significant effects of PME on the sleep behavior of newborns (17). Furthermore, the behavioral effects observed in the later life of these children were inconclusive and overall more sporadic (ebox 2). In eTable 3, all effects, both statistically significant and non-significant, are presented, including the direction of the effect.

eBOX 2. Effects of prenatal methamphetamine exposure (PME) on child behavior.

At the age of five years and 7.5 years, children with PME were more likely to show externalizing behavior (hyperkinetic disorders and disorders of social behavior; g = 1.625 [1.365; 1.884] and g = 0.255 [0.024; 0.486], respectively) as well as signs of ADHS (g = 0.167 [-0.065; 0.398] and g = 1.5 [1.246; 1.754], respectively) (22). In addition, after five years they were more likely to show emotionally reactive (g = 0.26 [0.04; 0.48]) or aggressive behavior (g = 0.265 [0.045; 0.485]) which persisted until the age of 7.5 years (g = 0.297 [0.066; 0.529]), an age at which they also broke rules more frequently (g = 0.302 [0.07; 0.533]) (11, 32). No somatic complaints were noted after five years and 7.5 years. While anxious and/or depressive behavior was still observed after five years, it did not persist thereafter (11, 32).

Results of the quality assessment

The following methodological shortcomings were considered in the summary assessment of study quality (eTabelle 2):

• Relevant confounding variables were not accounted for (10, 11, 14, 35, 37– 40), as defined by Smith et al. 2006 and 2008 (26, 27)

• Lack of objective measurement of exposure—that is, only self-reported MA use (35, 38)

• Inadequate duration of the follow-up (10, 17, 28, 33)

• Furthermore, statistical parameters, such as confidence intervals (18, 32) or criteria for the model’s quality, were missing at times (22, 23).

Further details on study quality are presented in eBox 3.

eBOX 3. Detailed results of the quality assessment.

In two studies, exposure was not measured objectively, but determined based on maternal self-report alone (35, 38). Likewise, in two studies, the respective outcome was not measured in such a way that a risk of bias can be ruled out (36, 39). The included studies showed weaknesses especially with regard to potential confounding variables. One-third of the studies did not consider the complete set of covariates relevant for the respective outcome (10– 12, 14, 18, 24, 35, 37– 39). Examples of missing covariates include: for the assessment of behavior after 6.5 years the relationship status of the parents or guardians (11), for the assessment of the development of growth over a period of three years the size of the infant at birth (12), and for the assessment of neonatal outcomes any information about the mother, apart from alcohol consumption (37). In two studies, the confounding variables were measured, but not included in the subsequent statistical analysis of the outcome (18, 39). In five of the 19 studies with a follow-up of more than five days postnatum, the information about differences between the sample at baseline and the sample at the time of the follow-up („loss to follow up”) was incomplete (10, 14, 17, 31, 33). Likewise, in five studies it remains unclear whether the follow-up was long enough to reliably determine the effects of prenatal methamphetamine exposure (PME) (10, 17, 28, 29, 33). Due to the missing loss-to-follow-up information, in particular, and the missing confounder information in some studies, the reliability of the study overall is lower (10, 14, 36– 38); however, this also applies to missing confidence intervals in effect sizes (18, 32) or missing information about the quality of the model in more complex regression analyses or structural equation models (23, 24).

The risks of bias for the included studies are thus as follows:

• Lack of objective measurement of exposure, i.e. only self-reported MA use information L (35, 38)

• Failure to take all relevant variables into account to prevent confounding, either in the study design or in the statistical analysis (35, 38– 40)

• No standardized collection of data on potential confounding variables, e.g. by means of valid questionnaires (35, 37– 40)

• No objective outcome measurement, i.e. lack of blinding of the study’s assessors with regard to exposure status (37, 39)

As the result of inconsistencies in the consideration of potential confounding variables, in particular, and of the different countries of origin of the studies and the related differences in health care systems, the study populations vary widely. In combination with differences in study design and samples size (etable 2), this is a major contributor to the heterogeneity of the studies included in the review.

Discussion

In recent years, the effects on PME on child development were investigated in various studies and are summarized in this review.

Based on case reports with cerebral imaging findings, effects of PME on the development of central brain areas, such as the superior and posterior corona radiata, could already be assumed (e1). The results of the analyzed cohort studies illustrate the clinical relevance of PME. For example, children with PME had a smaller head circumference at birth and during the first years of life; in addition, they were more commonly affected by neurocognitive problems. Since not all studies adequately adjusted for additional tobacco and alcohol consumption as well as low socioeconomic status—which have a negative impact on long-term development—, it is not possible to attribute the effect on PME alone in all cases. Similar effects have already been described for prenatal exposure to other drugs (e2). Newborns had a smaller head circumference after exposure to methadone and marihuana, while exposure to cocaine and methadone affected cognitive and motor development (e3). With regard to functional consequences of PME on child behavior, the available studies are inconclusive: At the age of five years, mostly an internalizing behavior was observed, whereas at the age of seven years externalizing behavior was predominant (eTable 3).

Implications for prevention and management

When the setting in which a child grew up was considered in the analyses, it was found that a social environment characterized by quarrelling, violence, poverty, and low educational attainment of the primary caregivers was associated with a more pronounced manifestation of the effect of PME on behavioral development later in life (e4), even after adjustment for socioeconomic status and educational attainment, as in the IDEAL publications (e5). This is relevant given that many of the affected mothers are single moms, have a lower social status and are more inclined to depression or lack of perspective (e6).

In order to prevent MA consumption during pregnancy and to ensure that children with PME grow up in a stable social environment, withdrawal therapy with the goal of maternal abstinence should be started as early as possible in pregnancy (e7) and include interventions addressing individual reasons for ME use (e8). From the study data, risk profiles (e.g. social status and depression) can be derived and used to approach expectant mothers (e9). Multi-professional care networks can help identify drug use in pregnant women early on and convince them of the need for screening. (e10), as this is the ideal setting for withdrawal interventions (e11). Initial results of an evaluation of an integrated care system for mother-child pairs with PME have shown that a bond between mother and child can be established in a perinatal center (e11). However, studies evaluating the effectiveness of an MA withdrawal program during pregnancy and guidelines on how to conduct such a program are missing (e12).

Methodological strengths and limitations

This registered systematic review was assessed for completeness using the AMSTAR check list. In addition, the PRISMA checklist was used for the preparation and reporting of this review (e13). Yet, there are still some methodological limitations of this review which should be highlighted. Not all studies established the exposure to MA in an objective, quantified manner. Furthermore, specific effects of PME were not detected in all IDEAL cohorts to the same extent. Some of the methodological limitations (ebox 4) resulted in high heterogeneity (I²) on the calculated total effects of PME (figure 3).

eBOX 4. The modifying effect of family and socioeconomic background.

Among children with prenatal methamphetamine exposure (PME) who still lived with their natural mother when they were five years old (n = 56), behavioral problems, as identified by overall scores of the Child Behavior Checklist (n = 21 versus n = 10), and interaction with youth welfare offices (n = 11 versus n = 4) where more common compared to those children with PME who were taken into care or grew up in foster families (n = 41) (30). As the result of extreme poverty of the parents/guardians, a change of the parents/guardians as the primary caregivers of the child and an overall lower social status, among others, children with PME rather showed externalizing behavior (b = 0.83, p < 0.05; R² = 0.11), even if these variables were considered in the matching (32). Furthermore, externalizing behavior of five-year-old children with PME was less expressed in a poorer neighborhood (odds ratio [OR] = 0.903; p = 0.036) and more expressed if the parents/guardians had psychological problems (OR = 2.794, p = 0.008) and experienced a higher level of parenting stress (OR = 6.615, p = 0.016) (30). Behavioral control of five-year-old children with PME was also affected by an unstable living environment plagued by conflict, which in turn was associated with poorer executive function at the age of 6.5 years. (beta = 0.21. p < 0.01) (11).

Persistence of maternal methamphetamine use after birth of the child was associated with uninhibited behavior of the child at the age of 6.5 years (beta = 0.227, p = 0.003). Additional maternal tobacco use after birth had a comparable effect (beta = 0.273, p = 0.001) (18) and was significantly associated with decreased reaction times of children with PME. The same is true for psychological problems of the parents/guardians as well as experiences of physical and sexual violence up to the age of 5.5 years, even if no effect sizes were reported (15). All reported associations were independent of socioeconomic status and they were more noticeable compared to children without PME growing up under comparable conditions (11, 15, 30, 32). Even though children with PME were more likely to grow up in precarious conditions across the studies considered in this review (11, 15, 30, 32)—especially if maternal ME use persisted after birth (18)—, no cause-effect relationship between social environment and PME can be established.

Supplementary Material

eMethods 1

1. Summary of search strategy

2. Criteria for data extraction

3. Studies excluded during full-text screening, with reasons (etable 4)

4. Effect size calculation

Hedges‘ g was used to calculate effect sizes for both individual effects and as the basis for calculating pooled effect sizes in the forest plots. This measure is suitable for reporting effect sizes if the sizes of the samples to be compared are not identical, since the calculation involves weighting for sample size. For the pooled analyses (meta-analyses), the standardized mean difference (SMD) measure was calculated using a random-effects model since there were different populations in which exposure was measured differently. This, on the one hand, did preclude the use of a fixed-effects model and, on the other hand, made it necessary to standardize the pooled-effects estimator (SMD). The weighting methods used in the meta-analyses were those described by Hartung, Knapp, Sidik, and Jonkman, achieving more precise effect estimates with a smaller number of studies.

• Databases searched: Embase, Medline, PscyInfo

• Manual search: journals not listed in the above mentioned databases (“Sucht”), clinical practice (S3) guideline “Methamphetamine-related Disorders” and references of included studies

• Search period: 1 January 1990 to 30 November 2017, updated in November 2019

• Screening: two independent reviewers (FH and SD), involvement of a third reviewer in case of disagreement (MR)

• Data extraction: two independent reviewers (FH and SD), checked by a third reviewer (LH)

• Population (for example, sample size, age, response rate, duration of follow-up, matching of exposure group and control group, confounders considered)

• Measurement of exposure

• Somatic, psychological, psychosocial, and developmental outcomes (including measuring method)

• Key findings on effects of prenatal methamphetamine exposure (PME)

• Assessment of the methodological quality of the studies

eFigure 2.

The Apgar score at one minute was lower in newborns with PME.

SMD, standardized mean difference; 95% CI, 95% confidence interval

eFigure 3.

For the Apgar score at five minutes for newborns with PME, the results of the studies were inconsistent.

SMD, standardized mean difference; 95% CI, 95% confidence interval

eTable 4. Studies excluded during full-text screening, with reasons.

Bibliographic data	Reason for exclusion
Chang L, Oishi K, Skranes J, et al.: Sex-specific alterations of white matter developmental trajectories in infants with prenatal exposure to methamphetamine and tobacco. JAMA Psychiatry 2016; 73: 1217–27.	Inadequate outcome: neuroimaging study
Derauf C, LaGasse L, Smith L, et al.: Infant temperament and high-risk environment relate to behavior problems and language in toddlers. J Dev Behav Pediatr 2011; 32: 125–35.	No control group
Dinger J, Näther N, Wimberger P, et al.: Steigender Konsum von Crystal Meth in Sachsen und dessen Risiken für Mutter und Kind – Erfahrungen an einem Perinatalzentrum Level I aus pädiatrischer Sicht. Z Geburtshilfe Neonatol 2017; 221: 73–80.	No control group
Elliott L, Loomis D, Lottritz L, et al.: Case-control study of a gastroschisis cluster in Nevada. Arch Pediatr Adolesc Med 2009; 163: 1000–6.	Ineligible population: mothers
Forrester MB, Merz RD: Risk of selected birth defects with prenatal illicit drug use, Hawaii, 1986–2002. J Toxicol Environ Health, Part A. 2007; 70: 7–18.	No control group
Hon KL, Chan MHM, Ng MHJ, et al.: Urine comprehensive drug screen, low birth weight and withdrawal symptoms in a neonatal unit: a case control study. Curr Clin Pharmacol 2016; 11: 274–81.	No control group
Kirlic N, Newman E, LaGasse LL, et al.: Cortisol reactivity in two-year-old children prenatally exposed to methamphetamine. J Stud Alcohol Drugs 2013; 74: 447–51.	No control group