The association between maternal tobacco smoking during pregnancy and the risk of attention-deficit/hyperactivity disorder (ADHD) in offspring: A systematic review and meta-analysis

Abstract

Introduction

Maternal tobacco smoking during pregnancy is a significant public health concern with potential long-lasting effects on child development. ADHD, a neurodevelopmental disorder characterized by inattention, hyperactivity, and impulsivity, may be influenced by prenatal nicotine exposure. This systematic review and meta-analysis examine the association between maternal tobacco smoking during pregnancy and the risk of ADHD in offspring.

Methods

Following PRISMA guidelines, we searched databases including PubMed, Web of Science, Cochrane Central, Embase, Scopus, CINAHL, LILACS, SciELO, Allied and Complementary Medicine Database (AMED), ERIC, CNKI, HTA Database, Dialnet, EBSCO, LENS, and Google Scholar for studies up to November 1, 2024. We included peer-reviewed studies reporting quantitative effect size estimates for the association between maternal tobacco smoking and ADHD. Study quality was assessed using the Newcastle-Ottawa Scale (NOS).

Results

We identified 2,981 articles and included 55 studies (4,016,522 participants) in the analysis. The meta-analysis showed a significant association between maternal tobacco smoking during pregnancy and increased risk of ADHD in offspring (pooled Odds Ratio (OR) = 1.71, 95% CI: 1.55-1.88; P < 0.001). Egger’s test indicated no publication bias (p = 0.204), but Begg’s test did (p = 0.042). By employing the trim and fill method, the revised OR was estimated to be 1.54 (95% CI: 1.40–1.70; P < 0.001). The OR were 2.37 (95% CI: 1.72–3.28; P < 0.001) in cross-sectional studies, 1.72 (95% CI: 1.49–2.00; P < 0.001) in case-control studies, and 1.53 (95% CI: 1.34–1.74; P < 0.001) in cohort studies. Meta-regression showed study design and study region significantly influenced heterogeneity (P < 0.10). Sensitivity and subgroup analyses confirmed the robustness of these findings.

Conclusion

This systematic review and meta-analysis demonstrate a significant association between maternal tobacco smoking during pregnancy and increased odds of ADHD in offspring. These findings highlight the need for prenatal care guidelines and tobacco smoking cessation programs for pregnant women to reduce ADHD risk and promote optimal neurodevelopmental outcomes. Future research should explore underlying mechanisms and potential confounders further.

Introduction

Maternal tobacco smoking during pregnancy remains a pervasive public health issue worldwide, with significant implications for maternal and child health [1,2]. The prenatal period is a critical window of development where exposure to harmful substances, such as those found in tobacco smoke, can have lasting effects on the developing fetus [3–5]. One of the neurodevelopmental disorders of particular concern in this context is Attention-Deficit/Hyperactivity Disorder (ADHD) [6]. ADHD is characterized by patterns of inattention, hyperactivity, and impulsivity that are inconsistent with developmental levels and significantly impair functioning across various domains of life [7]. Globally, ADHD affects approximately 2–7% of children and adolescents, with prevalence varying depending on diagnostic criteria, age, and location [8]. ADHD significantly impacts individuals, families, and society [7]. Children with ADHD often struggle with academic performance, social interactions, and behavior management. These challenges frequently continue into adulthood, affecting job success and personal relationships [9,10]. The economic impact of ADHD is substantial, encompassing healthcare expenses, productivity losses, and special education services. The annual societal cost of ADHD in childhood and adolescence is conservatively estimated to be $42.5 billion, with a range of $36 billion to $52.4 billion [11].

The etiology of ADHD involves genetic and environmental factors, including prenatal tobacco smoke exposure through maternal smoking [12,13]. Nicotine and other toxic substances in tobacco can cross the placenta and disrupt fetal brain development [14]. Studies have shown that nicotine affects neurotransmitter systems, particularly dopamine and norepinephrine [15–17], and alters neural circuits related to attention and self-regulation [18,19]. Additionally, nicotine induces oxidative stress and inflammation, impacting brain development [20].

Emerging research highlights the role of epigenetic mechanisms in mediating the association between maternal tobacco smoking and ADHD risk. Studies suggest that tobacco smoke exposure can alter DNA methylation patterns, potentially modifying gene expression related to neurodevelopment and increasing the risk of various psychiatric disorders, including ADHD [21–23]. These epigenetic modifications can disrupt the precise regulation of gene expression necessary for typical brain development, potentially contributing to ADHD symptoms [24]. This epigenetic perspective provides a crucial framework for understanding the enduring impact of prenatal nicotine exposure.

Epidemiological studies investigating the association between maternal tobacco smoking during pregnancy and ADHD in offspring have yielded diverse results [25–30]. Some large, well-designed observational studies report a positive correlation, finding that children exposed to tobacco smoke in utero through maternal tobacco smoking during pregnancy face an elevated risk of subsequently receiving an ADHD diagnosis in childhood [29,31–34]. These findings are supported by experimental research conducted on animals and human cells, which provide biological plausibility for the neurotoxic and disruptive effects of nicotine on fetal brain development pathways that are believed to be associated with ADHD [15,16].

However, not all epidemiological studies have found a statistically significant association between prenatal smoking exposure and ADHD [35,36]. Some smaller or earlier studies did not find a relationship or attributed the observed associations mainly to potential confounding factors. These factors include familial, genetic, and psychosocial factors that are correlated with both maternal tobacco smoking and the risk of ADHD. Examples of such factors are parental ADHD, low socioeconomic status, inadequate prenatal care, and other environmental factors [28,35–38]. The inconsistencies in study findings have created ongoing uncertainty regarding the true nature and magnitude of the potential effect.

Given the significant public health implications of clarifying this relationship, it is essential to conduct a rigorous evaluation and synthesis of the totality of evidence from the existing epidemiological literature. A systematic review and meta-analysis provide a robust methodological framework to quantitatively integrate data from multiple observational studies, critically appraise the quality of included research, and estimate the overall effect size while accounting for heterogeneity. This approach offers the best available means to address the knowledge gaps and inform clinical guidance and public health policy on this important issue [39,40].

In this systematic review and meta-analysis, we aim to investigate the association between maternal tobacco smoking during pregnancy and the risk of ADHD in offspring. We have three objectives: (1) to synthesize the existing epidemiological evidence on this association(2) to assess the influence of confounding factors and study quality on reported findings, and (3) to offer evidence-based insights that can guide clinical practice and public health policies aimed at reducing prenatal exposure to tobacco smoke and minimizing the risk of ADHD.

While previous meta-analyses have examined this association [1,2], our study expands upon prior work by encompassing a broader and more comprehensive search strategy across a wider range of databases, enabling the inclusion of a more complete set of relevant studies. Furthermore, we will employ advanced statistical techniques, including meta-regression and subgroup analyses stratified by factors such as ADHD diagnostic criteria and tobacco smoking ascertainment methods, to explore potential sources of heterogeneity and provide a more nuanced understanding of this complex relationship. By addressing these limitations of previous reviews, our study will offer a more robust and comprehensive evaluation of the impact of maternal tobacco smoking on ADHD risk, with implications for prenatal care guidelines, tobacco smoking cessation interventions, and public health initiatives promoting optimal neurodevelopmental outcomes. Ultimately, a clearer understanding of this association is essential for developing targeted interventions to improve child health and well-being and reduce the societal burden of ADHD.

Materials and methods

Study design and search strategy

This systematic review and meta-analysis were designed to examine the association between maternal direct tobacco smoking during pregnancy and the risk of ADHD in children. We adhered strictly to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to ensure a transparent and reproducible methodology [41,42].

Comprehensive literature search.

An exhaustive literature search was conducted across multiple electronic databases, including PubMed, Web of Science, Cochrane Central, Embase, Scopus, CINAHL, LILACS, SciELO, Allied and Complementary Medicine Database (AMED), ERIC, CNKI, HTA Database, Dialnet, EBSCO, LENS, and Google Scholar. The search covered all publications from the inception of these databases up to November 1, 2024. This extensive time frame was selected to encompass the most recent and relevant research available.

To identify pertinent studies, we employed a combination of Medical Subject Headings (MeSH) terms and keywords related to maternal tobacco smoking, pregnancy, and ADHD. These terms included “Maternal tobacco smoking”, “pregnancy”, “prenatal”, “ADHD”, “Attention-deficit/hyperactivity disorder”, and their synonyms. Boolean operators (AND, OR) were used to combine these terms, ensuring a comprehensive retrieval of relevant articles.

Manual search and reference checking.

In addition to the electronic database search, we manually reviewed the reference lists of all included studies to identify any additional articles that may have been missed. This backward snowballing technique ensured a thorough capture of relevant literature.

Inclusion and exclusion criteria

Eligibility criteria.

To ensure the inclusion of high-quality studies, we established strict eligibility criteria. We included original research articles that explored the relationship between maternal tobacco smoking during pregnancy and ADHD in children. The studies had to be published in peer-reviewed journals to ensure the reliability and comprehensibility of the findings.

We considered various study designs, including case-control, cross-sectional, and retrospective/prospective cohort studies. These designs are well-suited for examining associations between exposure (maternal tobacco smoking) and outcomes (ADHD in children).

Eligible studies were required to report quantitative effect size estimates, such as odds ratios (ORs), hazard ratios (HRs), or relative risks (RRs), along with their corresponding 95% confidence intervals (CIs). This information was crucial for the meta-analysis to objectively assess the strength and precision of the association between maternal tobacco smoking and ADHD.

Exclusion criteria.

We excluded studies that did not involve human subjects, those with insufficient data, and articles such as reviews, editorials, and case reports. Additionally, we excluded studies that did not adequately measure the exposure or the outcome variables, as well as those with incomplete or unclear data reporting. Studies that failed to provide sufficient effect estimates or raw data for quantitative analysis were also excluded.

This review focuses specifically on the effects of maternal tobacco smoking during pregnancy. Studies examining other forms of tobacco use (e.g., chewing tobacco, snuff) or the effects of other smoked substances besides tobacco (e.g., cannabis) were excluded.

Study selection process

Initial screening.

The study selection process involved multiple stages to ensure the inclusion of relevant studies. Two independent reviewers initially screened the titles and abstracts of all identified articles. Articles that were clearly irrelevant to the study aims were excluded at this stage.

Full-text review.

The full texts were obtained and reviewed against the predefined inclusion and exclusion criteria for articles that appeared relevant based on their titles and abstracts. Both reviewers independently assessed each article to determine its eligibility.

Consensus and discrepancy resolution.

In cases where discrepancies arose between the reviewers during any stage of the screening process, a third independent reviewer was consulted. The third reviewer objectively examined the full texts in question and facilitated open discussion between all reviewers until a consensus decision was reached on final study selection. This systematic and rigorous screening methodology minimized potential bias or errors and ensured a robust, credible, and reproducible process.

Data extraction

Development of data extraction form.

To maintain the integrity and consistency of the data, a detailed standardized data extraction form was developed prior to conducting the extractions. This form underwent thorough pilot testing by the research team to ensure it effectively captured all essential study details in an organized and reproducible manner.

Data extraction process.

Two independent reviewers utilized this well-defined form to systematically extract relevant information from each included study. The extracted data from each study underwent rigorous cross-checking by both reviewers to ensure accuracy. Any discrepancies were resolved through open discussion and consensus decision-making, involving a third reviewer when necessary.

The comprehensive data extraction form enabled the collection of important bibliographic details, study characteristics, population descriptors, exposure and outcome assessment methods, reported effect estimates, adjusted covariates, and quality assessment ratings. By capturing this information in a standardized manner from each study, consistency and completeness were ensured.

Quality assessment

Newcastle-Ottawa Scale (NOS).

The methodological quality of the included studies was carefully assessed using the widely accepted Newcastle-Ottawa Scale (NOS) [43]. This scale evaluates non-randomized studies based on three key domains: selection of study groups, comparability between groups, and ascertainment of exposure and outcome variables. Each study was assigned a detailed quality rating on a scale ranging from zero to nine points, with higher scores indicating stronger methodological conduct and reporting. Scores below 5 indicated low-quality articles, scores between 5 and 7 indicated moderate quality, and scores of 8 or higher indicated good quality [44,45].

Independent quality assessment.

Two independent reviewers applied the standardized NOS criteria to assess the quality of each study. Any discrepancies in scoring were resolved through open discussion, involving a third reviewer if necessary. Based on the total points achieved, studies were categorized as having good, moderate, or low methodological quality.

The NOS takes into account important criteria specific to case-control and cohort study designs, such as representative case selection, well-defined controls, comparability of cases and controls on key confounding factors, and robust ascertainment of exposure status. This rigorous and standardized approach to quality assessment allowed for objective evaluation of potential biases within and between studies. It also facilitated meaningful subgroup analyses to explore whether the strength of associations varied based on study quality ratings.

Study registry.

The protocol of this study has been registered in PROSPERO with code CRD42024595682. The study protocol was approved by the Ethics Committee of Shahrekord University of Medical Sciences (IR.SKUMS.REC.1403.113).

Statistical analysis

Meta-analysis.

To ensure the validity and reliability of our systematic review and meta-analysis findings, we employed rigorous statistical and graphical techniques to assess heterogeneity comprehensively. For studies that reported effect estimates separately for different time periods of exposure, we conducted meta-analyses methods to synthesize the stratified estimates into overall effects within each study. This approach maximized the inclusion of data without duplicating participant populations. Similarly, if studies provided results stratified by important covariates, but did not report an overall estimate, we performed meta-analyses to combine the stratified effects. In cases where studies presented raw exposure and outcome group data without a calculated effect size, we used Stata software to generate OR estimates with 95% confidence intervals.

Heterogeneity assessment.

To assess between-study heterogeneity, we utilized both statistical tests and visual inspection of forest plots. The Chi-square test helped determine if observed differences were due to chance alone, with a significance level of P < 0.10 indicating statistically significant heterogeneity. Additionally, we calculated the I² statistic to quantify the percentage of total variation attributed to heterogeneity rather than sampling error. If significant heterogeneity was detected, we selected random-effects models for meta-analyses [46,47]. We carefully examined forest plots to visually assess the overlap and distribution of confidence intervals across studies. Any potential outliers were further investigated through meta-regression model, subgroup analyses, and sensitivity analyses to identify potential sources of heterogeneity.

Exploration of heterogeneity

Meta-regression.

To explore the impact of covariates on heterogeneity, we conducted univariate and multivariate meta-regression using Stata software. Covariates such as study design, study year, sample size, study quality assessment score, methods of ascertaining tobacco smoking, methods of ascertaining ADHD, and geographical area were examined [48].

Sensitivity analysis.

To evaluate the robustness of our findings, we conducted sensitivity analyses by systematically excluding each study one at a time and re-running the meta-analysis. This process helped to determine if any particular study had a disproportionate influence on the overall results [49].

Subgroup analysis.

Subgroup analyses were performed to investigate potential sources of heterogeneity and to explore whether the association between maternal tobacco smoking during pregnancy and ADHD varied across different study characteristics. Subgroups were defined based on study design, study year, sample size, study quality assessment score, methods of ascertaining tobacco smoking, methods of ascertaining ADHD, and geographical area. These analyses provided deeper insights into the relationship and allowed for more nuanced interpretations of the findings [50].

Assessment of publication bias.

We assessed publication bias using both graphical and statistical methods. Funnel plots were visually inspected for asymmetry, which can indicate the presence of publication bias. Additionally, Egger’s regression test and Begg’s adjusted rank correlation test were conducted to statistically evaluate the likelihood of publication bias. To address the potential bias from missing unpublished studies, we used the trim and fill method. These methods provided a comprehensive assessment of the potential impact of publication bias on our meta-analysis findings [47,51].

Missing data.

In the course of our analyses, we encountered several variables that had missing data. To maintain the integrity and accuracy of our findings, we made the decision to exclude these variables from specific analyses that necessitated complete datasets. This exclusion was particularly relevant for more complex statistical techniques, such as meta-regression and subgroup analyses, which require fully populated datasets to yield reliable and interpretable results.

Software.

All data analyses were performed using Stata 17 software, ensuring rigorous and reliable synthesis of the evidence [52].

Results

Characteristics of included studies

An extensive electronic search using specific keywords identified 2,981 articles. After removing 1,203 duplicates, 1,778 articles remained for further evaluation. These articles were screened based on predefined criteria, leading to the exclusion of 1,697 articles. Consequently, 81 relevant articles were identified, of which 13 were excluded for not reporting effect sizes or the inability to calculate them, 5 study focused on the relationship between parental smoking and ADHD in offspring, 6 studies evaluated exposure to second-hand smoke, and 3 articles were duplicates based on the same dataset. This rigorous selection process resulted in 54 articles for the study. An additional study was identified through reference checking, bringing the total to 55 articles for the systematic review and meta-analysis [25–38,53–93] (Fig 1).

Fig 1. Flowchart of Study Selection for Meta-Analysis.

A total of 55 studies, conducted between 1998 and 2024 across various countries including the United States, Finland, Sweden, Brazil, the Netherlands, Japan, the UK, Spain, China, Australia, New Zealand, Norway, Canada, France, Sweden, South Korea, Turkey, Romania, Bulgaria, Lithuania, Germany, Denmark, Egypt, and India, were analyzed to examine the association between maternal tobacco smoking during pregnancy and the risk of ADHD in offspring. These studies collectively included 4,016,522 participants [25–38,53–93] (Tables 1–3).

Table 1. Characteristics of Studies Included in the Meta-Analysis.

First author	Year	Country	Study design	Sample size	Participants with ADHD	Controls	N of tobacco smoking mothers	N of non- tobacco smoking mothers	Gender		Average offspring age (Year)	Average mother age(year)	NOS score
									Boy	Giral
Milberger S [53]	1998	US	Case-Control	260	140	120	NA	NA	158	102	6-17	NA	4
Mick E [54]	2002	US	Case-Control	522	280	242	60	462	260	262	6-17	29.9	5
Kotimaa AJ [94]	2003	Finland	Cohort	8,478	808	7,670	2,428	6,050	4,304	4,174	8	27	7
Knopik VS [56]	2005	US	Cohort	3,872	255	3,617	1,429	2,443	NA	NA	14.4	NA	6
Rodriguez A [57]	2005	Sweden	Cohort	414	NA	NA	NA	NA	142	146	NA	27	5
Flick LH [59]	2006	USA	Cohort	733	21	712	246	487	NA	NA	NA	22.9	5
Braun JM [58]	2006	USA	Cross-sectional	4,704	135	4,569	616	4,014	2,264	2,440	NA	NA	6
Schmitz M [95]	2006	Brazil	Case-Control	200	100	100	48	152	136	64	6-18	39	4
Wakschlag LS [60]	2006	US	Cohort	448	78	370	166	282	448	0	7	NA	5
Nigg JT [61]	2007	US	Case-Control	713	94	619	247	466	NA	NA	6-11	27.4	5
Yoshimasu K [96]	2009	Japan	Case-Control	360	90	270	73	197	209	61	6-15	38.8	5
Froehlich TE [64]	2009	USA	Cross-sectional	2,588	222	2,365	NA	NA	NA	NA	NA	NA	6
Biederman J [63]	2009	USA	Case-Control	536	88	448	115	421	303	274	NA	NA	5
Altink ME [62]	2009	Netherlands	Case-Control	184	79	105	34	150	NA	NA	NA	NA	4
Motlagh MG [70]	2010	China	Case-Control	222	52	65	NA	NA	NA	NA	11.8	28.79	4
Nomura Y [71]	2010	US	Cohort	214	65	88	13	52	NA	NA	4.31	31.45	4
Anselmi L [66]	2010	Brazil	Cohort	4,423	880	3,546	NA	NA	NA	NA	NA	NA	6
Agrawal A [65]	2010	USA	Cross-sectional	1,122	141	1,201	159	963	NA	NA	NA	NA	6
Lindblad F [69]	2010	Sweden	Cohort	982,856	6,496	976,360	187,106	795,750	499,231	483,625	6-19	30	8
Hutchinson J [68]	2010	UK	Cohort	13,654	1,199	12,455	3,166	10,488	7,329	6,325	3	27	7
Ball SW [67]	2010	US	Cohort	2,024	219	1,805	1,212	812	943	1,081	7	NA	6
Koshy G [73]	2011	England	Cross-sectional	945	32	913	267	688	465	480	7.28	NA	6
Sciberras E [97]	2011	Australia	Cohort	3,474	64	3,410	NA	NA	1,771	1,703	6-7	34.9	6
Gustafsson P [72]	2011	Sweden	Case-Control	32,012	237	31,775	8,303	22,675	NA	NA	NA	NA	7
Obel C [74]	2011	Finland	Cross-sectional	868,449	7,023	861,426	135,275	733,174	443,076	425,373	NA	NA	8
Langley K [77]	2012	UK	Cohort	5,637	121	5,516	1,014	4,623	NA	NA	7.6	NA	7
Ellis LC [76]	2012	Norway	Cross-sectional	777	34	743	116	661	449	474	NA	NA	6
Sagiv SK [37]	2013	US	Cohort	601	75	526	166	435	308	292	8	25	5
Thakur GA [79]	2013	Canada	Cross-sectional	436	NA	NA	NA	NA	356	80	NA	NA	5
Jaspers M [78]	2013	Netherlands	Cohort	1,816	419	1,397	523	1,293	1,060	756	1.5-4	NA	6
Zhu JL [87]	2014	Denmark	Cohort	53,848	1,056	52,792	4,776	49,072	27,645	26,203	7	29.5	7
Silva D [80]	2014	Australia	Case-Control	43,062	12,991	30,071	1,070	41,992	33,221	9,841	≤25	27	8
Skoglund C [81]	2014	Sweden	Cohort	813,030	NA	NA	129,827	638,400	392,865	375,282	NA	NA	8
Kovess V [98]	2015	Turkey, Romania, Bulgaria, Lithuania, Germany, and the Netherlands	Cross-sectional	3,769	494	3,275	807	2,962	NA	NA	NA	NA	6
Melchior M [36]	2015	France	Cohort	921	176	745	144	777	488	433	5	NA	6
Han J-Y [99]	2015	South Korea	Cross-sectional	19,940	1,769	18,171	14,493	5,447	9,855	68	6-18	39	7
Obel C [31]	2016	Denmark	Cohort	968,665	17,381	951,284	234,178	734,487	496,943	471,722	NA	28	8
Joelsson P [83]	2016	Finland	Case-Control	48,493	10,132	38,811	3,064	7,068	NA	NA	NA	NA	8
Oerlemans AM [88]	2016	The Netherlands	Cross-sectional	884	476	408	97	787	770	114	11.8	NA	9
Gustavson K [84]	2017	Norway	Cohort	93,258	2,035	102,262	7,930	85,355	NA	NA	NA	NA	8
Ezquiaga Echezarreta H [89]	2017	Spain	Cohort	323	8	315	68	255	NA	NA	4	NA	6
Schwenke E [85]	2018	Germany	Cohort	572	43	529	176	439	NA	NA	NA	30	5
Minatoya M [32]	2019	Japan	Cohort	2,150	649	1,501	268	1,882	1,076	1,074	NA	33	7
Lin L-Z [25]	2021	China	Cohort	48,612	2,170	43,392	286	45,276	22,657	22,905	11	NA	8
Miyake K [33]	2021	Japan	Cohort	1,150	112	609	109	612	579	571	NA	31.55	6
Huhdanpää H [90]	2021	Finland	Cohort	697	23	674	33	664	366	333	5	31.2	7
Garrison-Desany HM [26]	2022	USA	Cohort	3,138	486	2,652	120	460	1,583	1,555	NA	NA	6
Haan E [34]	2022	UK, Netherland and Norway	Cohort	NA	NA	NA	4,525	49,062	28,080	27,046	NA	NA	4
Howell MP [27]	2022	USA	Cohort	133	35	98	16	117	NA	NA	NA	NA	4
Liu D [28]	2022	China	Cohort	2,477	103	2,374	655	1,822	1,338	1,139	7	27.55	6
Younis EA [29]	2023	Egypt	Cross-sectional	1,048	110	938	72	968	586	462	4.5	NA	6
Xiaomei F [91]	2023	China	Cross-sectional	3,122	196	2,926	–	–	–	–	7	31	6
Li Q [30]	2024	China	Cross-sectional	8,086	850	7,236	NA	4,623	4,623	3,463	5	NA	8
Nielsen TC [92]	2024	Australia/New Zealand	Cohort	908,770	16,297	892,473	119,959	788,811	469,834	438,936	7	NA	9
Lebeña A [93]	2024	Sweden	Cohort	16,365	755	15,610	1,705	14,660	8,910	7,455	14.4	30	9

Table 3. Adjusted Variables in Included Studies.

First author	year	Adjusted variables
Milberger S [53]	1998	Socioeconomic status, maternal and paternal IQ, maternal and paternal ADHD.
Mick E [54]	2002	Maternal age at birth, social adversity indicators, parental history of ADHD, conduct disorder (CD), antisocial personality disorder (ASPD), and comorbid CD.
Kotimaa AJ [94]	2003	Family structure, maternal age, socioeconomic status, and prenatal alcohol use.
Knopik VS [56]	2005	Birth weight, parental alcohol history, and maternal alcohol use during pregnancy.
Rodriguez A [57]	2005	Child’s sex.
Flick LH [59]	2006	Individual psychiatric diagnoses, diagnostic categories, and “any current disorder” diagnoses.
Braun JM [58]	2006	Race/ethnicity, sex, age, blood lead level, ferritin level, smoker presence in the home, preschool attendance, and insurance status.
Schmitz M [95]	2006	Alcohol use during pregnancy, birth weight, maternal ADHD, and oppositional defiant disorder.
Wakschlag LS [60]	2006	Drug use.
Nigg JT [61]	2007	Maternal substance uses disorders, birth weight, education level, and location.
Yoshimasu K [96]	2009	Paternal educational background, family income, family structure, parental mental health history (schizophrenia, depression, alcohol dependence, personality disorder), maternal stress, smoking (active and passive) and drinking habits during pregnancy, maternal ADHD tendency, severity of child’s ADHD, and iron-rich diet in children.
Froehlich TE [64]	2009	Current household smoke exposure, gender, age, race/ethnicity, income, preschool attendance, mother’s age at birth, and birth weight.
Biederman J [63]	2009	Maternal age at birth, social class, offspring age at baseline, offspring sex, parental history of ADHD and conduct disorder, prenatal exposure to alcohol or illicit drugs, study origin (Boys ADHD, Girls ADHD), and number of assessments.
Altink ME [62]	2009	Age, gender, IQ, birth weight, oppositional symptoms (Conners A-scale), anxious-shy symptoms (Conners D-scale), total parental ADHD symptoms, maternal age, and socioeconomic status.
Motlagh MG [70]	2010	Child’s sex, inattentive and hyperactivity scores, multiple pregnancy complications, and high coffee consumption (>3 cups in 24 hours).
Nomura Y [71]	2010	Age, gender, socioeconomic status, birth weight, race, self-reported maternal and paternal ADHD symptoms (Conners’ Adult ADHD Rating Scale), and maternal alcohol use during pregnancy.
Anselmi L [66]	2010	Skin color, family income, alcohol use during pregnancy, birth weight, gestational age, intrauterine growth restriction.
Agrawal A [65]	2010	Adjusted for ethnicity, parental education, and study design.
Lindblad F [69]	2010	Age, birth year, sex, county of residence, maternal age, birth order, maternal education, single parenthood, social assistance, parental psychiatric/addictive disorders, small for gestational age, and low Apgar score.
Hutchinson J [68]	2010	Mother’s age at birth, number of children in the household, mother’s ethnicity, and psychosocial and parenting factors.
Ball SW [67]	2010	Source of ascertainment, family psychopathology, maternal education, and offspring sex.
Koshy G [73]	2011	Obesity, maternal smoking during pregnancy (heavy and light), household smoking, asthma diagnosis, preterm birth, and low birth weight.
Sciberras E [97]	2011	Child’s age and sex, maternal age at birth, birthplace (Australia/New Zealand), primary language at home, education status, income, biological parents’ presence, household size, marital status, maternal alcohol use during pregnancy, post-natal depression, and birth weight.
Gustafsson P [72]	2011	–
Obel C [74]	2011	Sex, birth year, maternal age, gestational age at birth, and parity.
Langley K [77]	2012	Child’s sex, ethnicity, multiple births (twins), maternal alcohol use during pregnancy, and social class.
Ellis LC [76]	2012	Parental anxiety, depression, personality disorders, drug abuse, and socioeconomic characteristics.
Sagiv SK [37]	2013	Maternal age at child’s birth, maternal and paternal education, maternal marital status and household income at school age, maternal alcohol use during pregnancy, illicit drug use before birth, maternal IQ, depression symptoms, HOME score, gestational age, sex, race, breastfeeding, school type, and number of siblings in the home.
Thakur GA [79]	2013	–
Jaspers M [78]	2013	Alcohol use during pregnancy, low birth weight, and birth defects.
Silva D [80]	2014	Maternal age, marital status, first pregnancy, threatened abortion, preterm labor, maternal UTI, preeclampsia, labor complications, delivery type, gestational weeks, and birth weight.
Skoglund C [81]	2014	Sex, birth year, maternal parity, maternal age at child birth, cohabitation with the father, maternal education, and country of birth.
Zhu JL [87]	2014	Maternal age, parity, alcohol intake during pregnancy, parental socioeconomic status, parental psychopathology, and child’s gender.
Kovess V [98]	2015	Child’s sex and age, maternal age, education level, psychological distress, employment, marital status, number of live births, and geographical region.
Melchior M [36]	2015	Child’s sex, premature birth, birth weight, breastfeeding duration, maternal age at birth, psychological difficulties during pregnancy, post-pregnancy maternal depression, alcohol use during pregnancy, smoking post-pregnancy, paternal smoking, study center, parental education, family income, parental separation, number of siblings, and negative life events.
Han J-Y [99]	2015	Gender, age, father’s education, mother’s age at childbirth, marital status, delivery complications, vaccination history, and family history of ADHD.
Obel C [31]	2016	Sex, birth year, parity, mother’s age.
Joelsson P [83]	2016	Parental psychiatric history, maternal substance use history, parental age at birth, maternal socioeconomic status, birth weight for gestational age, number of previous births, and maternal marital status.
Oerlemans AM [88]	2016	Family size, stress during pregnancy, tobacco use during pregnancy, low parental age, suboptimal condition at birth, maternal infections.
Gustavson K [84]	2017	Parental age and education, parental ADHD symptoms, maternal (pre-pregnancy) and paternal BMI, maternal alcohol use during pregnancy, parity, child’s birth year, and geographic region.
Ezquiaga Echezarreta H [89]	2017	Socioeconomic status, parental education level
Schwenke E [85]	2018	–
Minatoya M [32]	2019	Family income during pregnancy, maternal alcohol use during pregnancy, parity, paternal smoking during pregnancy, and child’s sex.
Lin L-Z [25]	2021	Child’s age and sex, preterm birth, low birth weight, parental education level, income, maternal age, current maternal smoking, prenatal maternal smoking, and prenatal alcohol consumption.
Miyake K [33]	2021	Age, height, pre-pregnancy weight, parity, first trimester alcohol consumption, and household income; birth weight, gestational age, and infant sex.
Huhdanpää H [90]	2021	–
Garrison-Desany HM [26]	2022	Maternal factors (race/ethnicity, age, education, marital status, pre-pregnancy BMI), household income, nulliparity, and child’s sex.
Haan E [34]	2022	Age, ethnicity, marital status, education, financial difficulties, mental health, parental ADHD, smoking, alcohol and caffeine consumption, offspring gender, and parity.
Howell MP [27]	2022	Infant sex, race, socioeconomic status, and current maternal depression.
Liu D [28]	2022	Maternal alcohol use before and during pregnancy, offspring gender, maternal age, depression status during pregnancy, paternal obesity, and monthly income per capita.
Younis EA [29]	2023	Age, gender, residence, family history of psychological and neurological symptoms, ADHD symptoms, maternal neurological/psychological symptoms, passive smoking, and drug use during pregnancy.
Xiaomei F [91]	2023	Gender, maternal education level, history of epilepsy, maternal mode of delivery, premature birth, asphyxia at birth, maternal smoking during pregnancy, feeding pattern, lead exposure, parental relationship, parental divorce, beat-and-scold education style, picky eating, and learning difficulties.
Li Q [30]	2024	–
Nielsen TC [92]	2024	Gender, age, year and season of birth, maternal age, socioeconomic disadvantage remoteness, maternal country of birth, maternal developmental or conduct disorder.
Lebeña A [93]	2024	Sex, Maternal age, Paternal age, Maternal education, Paternal education, non-Swedish mother, non-Swedish father, Household income, Serious life event, Breastfeeding duration.

Table 2. Odds Ratio of Maternal Tobacco Smoking During Pregnancy and ADHD in Offspring.

First author	Year	Ascertainment of ADHD	Ascertainment of tobacco smoking	Odds Ratio(95%CI)
Milberger S [53]	1998	Children were interviewed by a specialist in child and adolescent psychiatry	Diagnostic interview for children and adolescents–parent version	2.70 (1.10–7)
Mick E [54]	2002	Interviews with each parent and medical record	Collected by structured diagnostic interview and self-reported regarding her own psychiatric history	2.10 (1.10–4.10)
Kotimaa AJ [94]	2003	Parents and teachers completed questionnaires on the child’s development or behavior	Recorded at recruitment	1.30 (1.08–1.58)
Knopik VS [56]	2005	Interview with mother	Extracted from maternal interview data	1.25 (0.82–1.91)
Rodriguez A [57]	2005	Reported by mothers and teachers	Self-report	1.06 (1–1.13)
Flick LH [59]	2006	The Diagnostic Interview Schedule was used	Cigarettes smoked during pregnancy was obtained from birth certificates	2.74 (1.26, 5.96)
Braun JM [58]	2006	Based on the parents report of a diagnosis of ADHD and stimulant medication use	Based on the parent’s report	2.5 (1.2–5.2)
Schmitz M [95]	2006	Screened by SNAP-IV and diagnosed by child and adolescent psychiatrist	Collected by direct interview with the biological mother	2.01 (0.65–6.18)
Wakschlag LS [60]	2006	Interview with parents and children	Extract from maternal report on background interview	1.16 (0.69–1.94)
Nigg JT [61]	2007	Interview with mothers	Collected by direct interview with the biological mother	1.27 (0.85–1.90)
Yoshimasu K [96]	2009	Diagnosed by experienced psychiatrists or pediatricians	Investigated by a questionnaire	1.30 (0.50–3.60)
Froehlich TE [64]	2009	The Diagnostic Interview was used	Mother self-report	2.4 (1.5–3.7)
Biederman J [63]	2009	In 3 stages: clinical diagnosis, a telephone questionnaire administered to the mother, a diagnostic assessment with a structured interview	First mothers were directly questioned, second was derived from the mother’s self-reported structured diagnostic interview	2.5 (1.39, 4.51)
Altink ME [62]	2009	Ascertained from data of the International Multi-center ADHD Gene project (clinical diagnosis)	Mother self-report	3.29 (1.48–7.30)
Motlagh MG [70]	2010	Parental report using the DuPaul-Barkley ADHD rating scale	Mothers were interviewed directly using a semi-structured scale	13.5 (1.6–113.2)
Nomura Y [71]	2010	The ADHD-Rating Scale-IV (ADHD-RS-IV; DuPaul et al., 1998) was distributed to parents	Self-report demographic questionnaire	4.00 (1.36–11.12)
Anselmi L [66]	2010	Mothers answered the Strengths and Difficulties Questionnaire in order to evaluate ADHD	Mothers answered “yes/no” as to whether they had smoked during the pregnancy	1.28 (1.13–1.45)
Agrawal A [65]	2010	Self-reported meeting criteria for ADHD by mothers	Self-reported whether mothers had smoked during pregnancy	1.53 [1.00,2.35]
Lindblad F [69]	2010	Swedish Prescribed Drug Register	Collected by the midwife at the first visit to the maternity health clinic, 8–12 weeks after conception	1.73 (1.46–2.05)
Hutchinson J [68]	2010	Interview with mothers	Collected by interview	1.60 (1.27–2)
Ball SW [67]	2010	Diagnosed by psychologists	Collected beginning with the initial prenatal assessment and prospectively up to the day of birth by interview	1.14 (0.90–1.44)
Koshy G [73]	2011	ADHD was determent by the question	Parent self-report	3.19 (1.08–9.49)
Sciberras E [97]	2011	Interview with the primary caregiver	Investigated by a questionnaire	3.31 (1.49–7.39)
Gustafsson P [72]	2011	Diagnosed at the department of child and adolescent psychiatry	Investigated by a questionnaire	1.35 (1.14–1.60)
Obel C [74]	2011	Finnish Hospital Discharge Register	Finnish Medical Birth Register	2.01 (1.27–2.12)
Langley K [77]	2012	Interview with parents	Collected at 18- and 32-weeks’ gestation by interview	1.72 (1.14–2.61)
Ellis LC [76]	2012	Interview with parents	Reported by the mothers during the interview	2.59 (1.5–4.34)
Sagiv SK [37]	2013	Pediatric medical records	Extract from a questionnaire administered 2 weeks after birth	1.22 (0.70–2.15)
Thakur GA [79]	2013	Clinical interviews of the child and at least one of the two parents by a child psychiatrist.	The Kinney Medical Gynecological Questionnaire	1.06 (1–1.13)
Jaspers M [78]	2013	Interview with parents	Preventive Child Healthcare files	1.36 (1.07–1.73)
Zhu JL [87]	2014	Interview with parents	Preventive Child Healthcare files	1.63 (1.36–1.94)
Silva D [80]	2014	Monitoring of Drugs of Dependence system	Records of midwife’s notification system	1.83 (1.53–2.19)
Skoglund C [81]	2014	Specialist psychiatrist after a clinical somatic and psychiatric evaluation	Parent self-report	2.17 (1.65–2.86)
Kovess V [98]	2015	Mothers reported probable ADHD	Mothers report on self- smoking patterns during the pregnancy period	1.44 (1.06–1.96)
Melchior M [36]	2015	Interview with mothers	Investigated by a questionnaire	1.57 (0.77–3.22)
Han J-Y [99]	2015	Parents completed the ADHD DuPaul Rating Scale	Parents answered the questions on exposure to environmental tobacco smoke	2.64 (1.45–4.80)
Obel C [31]	2016	Medical Birth Register or medication record	Medical Birth Register	2.01 (1.94–2.07)
Joelsson P [83]	2016	Based on hospital diagnoses	Ascertained by maternity clinic nurses during routine prenatal visits during the second trimester of pregnancy and documented in health records	1.75 (1.65–1.86)
Oerlemans AM [88]	2016	Interview with parents	Maternal report and medical records	2.12 (1.26–3.58)
Gustavson K [84]	2017	Obtained from the Norwegian Patient Registry, also mothers responded to 6 questions about the	Mothers reported on smoking during the current and previous pregnancies.	1.48 (1.30–1.68)
Ezquiaga Echezarreta H [89]	2017	ADHD DMS-IV questionnaire administered to the adult accompanying the child during the 4-year pediatric control visit	Questionnaire on smoking habits administered to mothers at week 32 of pregnancy, and measurement of urinary cotinine levels	1.20 (1.17–1.22)
Schwenke E [85]	2018	ADHD Screening Questionnaire	Maternal report and medical records	2.51 (1.26–4.94)
Minatoya M [32]	2019	ADHD Screening Questionnaire	Maternal report and medical records	1.67 (1.17, 2.39)
Lin L-Z [25]	2021	Asked parents to fill out both the symptom inventory scale of ADHD (SIS-ADHD) and the Conners Abbreviated Symptom Questionnaire (C-ASQ) to measure ADHD symptoms in all children.	Parent self-report questioner	2.08 (1.10–3.93)
Miyake K [33]	2021	The child’s mother mainly filled out the questionnaire, including the ADHD survey for the child.	Cotinine level measurement Plasma cotinine levels at the third trimester of pregnancy were measured using a highly sensitive enzyme-linked immunosorbent assay kit	1.89 (1.14, 3.15)
Huhdanpää H [90]	2021	ADHD Screening Questionnaire	Maternal report and medical records	1.79 (0.86–3.74)
Garrison-Desany HM [26]	2022	Physician diagnosis reported in the child’s electronic medical record	Complete questioner by mother	0.33 (0.10 to 0.55)
Haan E [34]	2022	Complete questioner	Parental self-reported prenatal smoking	1.11 (1.00–1.23)
Howell MP [27]	2022	ADHD symptoms were obtained from maternal report on the Child Behavior Check List	Maternal report and medical records	3.71 (0.71, 6.70)
Liu D [28]	2022	Complete questioner by parent	Parent self-report questioner	1.12 (0.61–2.08)
Younis EA [29]	2023	ADHD Screening Questionnaire	Complete questioner by mother	4.81 (2.80–8.28)
Xiaomei F [91]	2023	Self-made questionnaire	Complete questioner by mother	12.55 (6–26.07)
Li Q [30]	2024	Complete questioner by parent	Parent self-report questioner	2.07 (1.67–2.56)
Nielsen TC [92]	2024	ADHD was identified from stimulant prescription records in the NSW Pharmaceutical Drugs of Addiction System (PHDAS) database.	Maternal smoking was documented in the NSW Perinatal Data Collection during pregnancy.	1.89 (1.82, 1.96)
Lebeña A [93]	2024	Physician diagnosis reported in the child’s electronic medical record	Complete questioner by mother	1.51 (1.24–1.85)

Maternal tobacco smoking during pregnancy and the risk of ADHD in offspring

The systematic review comprised 13 cross-sectional, 11 case-control, and 31 cohort studies. To address potential heterogeneity, a random-effects model was used for the meta-analysis. The meta-analysis results indicated a significantly higher odds of ADHD in children whose mothers smoked during pregnancy compared to those who did not. The pooled estimate showed an OR of 1.71 (95% CI: 1.55-1.88; P <  0.001), suggesting a considerable increase in the likelihood of ADHD in children exposed to maternal tobacco smoking during pregnancy (Fig 2).

Fig 2. Association Between Maternal Tobacco Smoking During Pregnancy and the Risk of ADHD in Children.

Evaluation of publication bias

Egger’s test (p =  0.204) showed no evidence of publication bias for the association between maternal tobacco smoking during pregnancy and ADHD in children. However, Begg’s test (p =  0.042) indicated some publication bias (Fig 3).

Fig 3. Evaluation of Publication Bias in Meta-Analysis Studies.

Revised effect size

We estimated the effect size for potentially missing studies using the trim and fill method. Fig 4 illustrates this method, which estimated results for 12 missing studies. Considering these missing studies, the revised OR was 1.54 (95% CI: 1.40–1.70; P <  0.001), reinforcing the significant relationship between maternal tobacco smoking during pregnancy and ADHD in children (Fig 4).

Fig 4. Trim and Fill Method Results for Estimating Effect Size of Missing Studies.

Meta-regression

Meta-regression analysis examined variables such as sample size, study design, study location, time period, methods of ascertaining tobacco smoking, methods of ascertaining ADHD, and study quality assessed using the Newcastle-Ottawa scale. Only study design (cross-sectional, case-control, and cohort) and study location (USA and Canada, Europe, Asia, South America, Australia, Africa) showed a significant association with heterogeneity (P <  0.10) (Table 4).

Table 4. Meta-Regression Results for Studies Investigating the Association Between Maternal Tobacco Smoking During Pregnancy and ADHD in Children.

Meta-regression REML estimate of between-study variance % residual variation due to heterogeneity Proportion of between—study variance explained Joint test for all covariates With Knapp-Hartung modification					Number of obs =  55 Taue2 =  0.07974 I-suuared_res =  74.12% Adj R-squared =  25.52% Model F (7,47) =  2.38 Prob>F =  0.0364
logor	Coef.	Std. Err.	t	p>‖t‖	[95% Conf. Interval]
Study design	−0.191	0.067	−2.84	0.007	−0.327	−0.055
Region	0.122	0.052	2.32	0.025	0.016	0.228
Study year	0.002	0.131	0.02	0.985	−0.263	0.268
Sample size	3.36	2.17	1.55	0.128	−1.00	7.72
Tobacco smoking ascertainment	0.020	0.06	0.31	0.757	−0.113	0.154
ADHD ascertainment	−0.077	0.066	−1.18	0.246	−0.211	0.055
Nos	−0.025	0.058	−0.43	0.669	−0.141	0.091
_cons	0.977	0.384	2.54	0.015	0.202	1.751

Sensitivity analysis

Sensitivity analysis involved sequentially removing each study to assess the robustness of the meta-analysis results. The estimated OR remained stable, indicating the robustness of the findings (Fig 5 and Table 5).

Fig 5. Sensitivity Analysis for the Association Between Maternal Tobacco Smoking During Pregnancy and ADHD in Children.

Table 5. Sensitivity Analysis of the Association Between Maternal Tobacco Smoking During Pregnancy and ADHD in Children.

Number	Publication first author	Year	OR (95% CI)
1	Milberger S	1998	1.71 (1.54–1.88)
2	Mick E	2002	1.70 (1.54–1.88)
3	Kotimaa AJ	2003	1.72 (1.56–1.91)
4	Knopik VS	2005	1.72 (1.56–1.90)
5	Rodriguez A	2005	1.74 (1.57–1.92)
6	Wakschlag LS	2006	1.72 (1.56–1.91)
7	Flick LH	2006	1.70 (1.54–1.88)
8	Schmitz M	2006	1.71 (1.55–1.89)
9	Braun JM	2006	1.70 (1.54–1.88)
10	Nigg JT	2007	1.72 (1.56–1.90)
11	Biederman J	2009	1.70 (1.54–1.88)
12	Altink ME	2009	1.70 (1.54–1.88)
13	Froehlich TE	2009	1.70 (1.54–1.88)
14	Yoshimasu K	2009	1.71 (1.55–1.89)
15	Ball SW	2010	1.73 (1.56–1.92)
16	Agrawal A	2010	1.71 (1.55–1.90)
17	Hutchinson J	2010	1.71 (1.55–1.90)
18	Nomura Y	2010	1.70 (1.54–1.88)
19	Anselmi L	2010	1.73 (1.56–1.91)
20	Motlagh MG	2010	1.70 (1.54–1.88)
21	Lindblad F	2010	1.71 (1.55–1.89)
22	Sciberras E	2011	1.70 (1.54–1.88)
23	Gustafsson P	2011	1.72 (1.56–1.91)
24	Obel C	2011	1.70 (1.54–1.88)
25	Koshy G	2011	1.70 (1.54–1.88)
26	Langley K	2012	1.71 (1.55–1.89)
27	Ellis LC	2012	1.70 (1.54–1.88)
28	Thakur GA	2013	1.74 (1.57–1.92)
29	Sagiv SK	2013	1.72 (1.55–1.90)
30	Jaspers M	2013	1.72 (1.56–1.90)
31	Skoglund C	2014	1.70 (1.54–1.88)
32	Zhu JL	2014	1.71 (1.55–1.90)
33	Silva D	2014	1.71 (1.54–1.89)
34	Han J-Y	2015	1.70 (1.54–1.88)
35	Melchior M	2015	1.71 (1.55–1.89)
36	Kovess V	2015	1.72 (1.55–1.89)
37	Joelsson P	2016	1.71 (1.55–1.90)
38	Oerlemans AM	2016	1.71 (1.54–1.88)
39	Obel C	2016	1.68 (1.53–1.84)
40	Ezquiaga Echezarreta H	2017	1.73 (1.56–1.92)
41	Gustavson K	2017	1.72 (1.55–1.91)
42	Schwenke E	2018	1.70 (1.54–1.88)
43	Minatoya M	2019	1.71 (1.55–1.89)
44	Huhdanpää H	2021	1.71 (1.55–1.89)
45	Miyake K	2021	1.71 (1.55–1.89)
46	Lin L-Z	2021	1.71 (1.54–1.89)
47	Haan E	2022	1.73 (1.57–1.89)
48	Garrison-Desany HM	2022	1.74 (1.57–1.92)
49	Howell MP	2022	1.70 (1.54–1.88)
50	Liu D	2022	1.72 (1.56–1.90)
51	Younis EA	2023	1.68 (1.52–1.86)
52	Xiaomei F	2023	1.67 (1.51–1.84)
53	Li Q	2024	1.70 (1.54–1.88)
54	Nielsen TC	2024	1.71 (1.54–1.88)
55	Lebeña A	2024	1.70 (1.54–1.88)

Subgroup analysis

Subgroup analyses were performed to explore potential sources of heterogeneity, considering sample size, study design, geographic location, publication year, methods of ascertaining tobacco smoking, methods of ascertaining ADHD, and study quality assessed using the Newcastle-Ottawa Scale. Results remained relatively consistent across different study designs: cross-sectional studies (OR =  2.37, 95% CI: 1.72–3.28), case-control studies (OR =  1.72, 95% CI: 1.49–2.00), and cohort studies (OR =  1.53, 95% CI: 1.34–1.74). Geographic variation was observed, with ORs of 1.54 (95% CI: 1.25–1.91) in the US and Canada, 1.63 (95% CI: 1.42–1.88) in Europe, 2.32 (95% CI: 1.60–3.35) in Asia, 1.29 (95% CI: 1.14–1.46) in South America, 1.89 (95% CI: 1.82–1.96) in Australia, and 4.81 (95% CI: 2.80–8.27) in Africa. Studies published in 2010 or earlier showed an OR of 1.58 (95% CI: 1.36–1.83), compared to 1.77 (95% CI: 1.56–2.01) for studies published from 2011 onwards. Sample size also appeared to influence the results: studies with fewer than 2,000 participants had an OR of 1.47 (95% CI: 1.32–1.62), while those with 2,000 or more participants had an OR of 1.70 (95% CI: 1.56–1.86).

Subgroup analysis based on tobacco smoking ascertainment method revealed ORs of 1.65 (95% CI: 1.40–1.95) for self-report, 1.75 (95% CI: 1.51–2.04) for interview, 1.76 (95% CI: 1.47–2.10) for medical records, and 1.28 (95% CI: 1.11–1.49) for biological measurement. Similarly, ADHD ascertainment method yielded ORs of 1.92 (95% CI: 1.59–2.31) for clinical interview/diagnosis, 1.58 (95% CI: 1.42–1.75) for self-report by child/parent or teacher report, and 1.66 (95% CI: 1.41–1.97) for medical records/databases. Finally, studies of good quality had an OR of 1.76(95% CI: 1.63–1.91), moderate quality studies had an OR of 1.49(95% CI: 1.34–1.65), and low-quality studies had an OR of 2.70(95% CI: 1.42–5.14) (Table 6).

Table 6. Subgroup Analysis of the Association Between Maternal Tobacco Smoking During Pregnancy and ADHD in Children.

Characteristics		Number of studies	OR (95% CI)	P-value
Study design	Case-control	11	1.72 (1.49–2.00)	≤0.001
	Cohort	31	1.53 (1.34–1.74)	≤0.001
	Cross-sectional	13	2.37 (1.72–3.28)	≤0.001
Study location	USA and Canada	16	1.54 (1.25–1.91)	≤0.001
	Europe	24	1.63 (1.42–1.88)	≤0.001
	Asia	9	2.32 (1.60–3.35)	≤0.001
	South America	2	1.29 (1.14–1.46)	≤0.001
	Australia	3	1.89 (1.82–1.96)	≤0.001
	Africa	1	4.81 (2.80–8.27)	≤0.001
Time period	2010 and before	21	1.58 (1.36–1.83)	≤0.001
	2011 and later	34	1.77 (1.56–2.01)	≤0.001
Sample size	<2000	27	1.47 (1.32–1.62)	≤0.001
	≥2000	28	1.70 (1.56–1.86)	≤0.001
Tobacco smoking ascertainment	Self-Report	16	1.65 (1.40–1.95)	≤0.001
	Interview	25	1.75 (1.51–2.04)	≤0.001
	Medical Records	11	1.76 (1.47–2.10)	≤0.001
	Biological Measurement	3	1.28 (1.11–1.49)	0.001
ADHD ascertainment	Clinical Interview/Diagnosis	12	1.92 (1.59–2.31)	≤0.001
	Self-Report by Child/Parent or Teacher Report	31	1.58 (1.42–1.75)	≤0.001
	Medical Records/Databases	12	1.66 (1.41–1.97)	≤0.001
Quality assessment	Low quality	7	2.70 (1.42–5.14)	0.002
	Moderate quality	28	1.49 (1.34–1.65)	≤0.001
	Good quality	20	1.75 (1.62–1.89)	≤0.001

Discussion

This systematic review and meta-analysis of 55 studies, encompassing over four million participants, provides compelling evidence that maternal tobacco smoking during pregnancy significantly increases the odds of ADHD in children (OR =  1.71, 95% CI: 1.55–1.88). This finding, indicating a 71% increased likelihood of ADHD in children exposed to prenatal tobacco smoke, remained consistent across various study designs, geographic regions, and time periods. Our results align with those of Huang et al. [21], who reported a 60% increased risk of ADHD in children of mothers who smoked during pregnancy based on their meta-analysis of 20 studies. These findings underscore the critical need for public health interventions aimed at reducing tobacco smoking during pregnancy.

Several physiological mechanisms could explain the link between maternal tobacco smoking during pregnancy and a heightened risk of ADHD in offspring [100,101]. Nicotine, the primary psychoactive substance in tobacco, can readily cross the placental barrier, impacting fetal brain development directly [19]. Exposure to nicotine during critical periods of brain development can disrupt neurotransmitter systems, particularly the dopaminergic system, which is vital for regulating attention and behavior [4,19].

Smoking during pregnancy can also lead to chronic fetal hypoxia due to carbon monoxide and other harmful substances in tobacco smoke [102]. Hypoxia can impair oxygen delivery to the fetal brain, resulting in neurodevelopmental deficits. Prolonged hypoxia can cause structural changes in the brain, such as reduced cortical thickness and altered connectivity, which are associated with ADHD symptoms [103,104].

Additionally, maternal tobacco smoking can trigger an inflammatory response, resulting in the release of pro-inflammatory cytokines [105]. These cytokines can cross the placental barrier and affect fetal brain development, potentially leading to neuroinflammation [106]. Neuroinflammation is linked to various neuropsychiatric disorders, including ADHD, suggesting that prenatal exposure to inflammatory agents could contribute to ADHD development [107,108].

Epigenetic modifications may partially explain the observed association between maternal smoking and ADHD in this meta-analysis. Prenatal exposure to tobacco smoke has been shown to disrupt DNA methylation patterns, impacting gene expression crucial for neurodevelopment [109]. Such alterations could have cascading effects on brain development, potentially increasing the susceptibility to ADHD and other neuropsychiatric disorders [24]. While this study does not directly investigate epigenetic mechanisms, the findings align with the growing body of literature suggesting that epigenetic dysregulation plays a significant role in the etiology of ADHD following in-utero tobacco exposure. Further research exploring specific epigenetic markers and their functional consequences in the context of maternal smoking and ADHD is warranted to elucidate the precise biological pathways involved [110].

The meta-regression and subgroup analyses shed light on the factors that contribute to the observed heterogeneity in the results of included studies [21,48,50]. The study design was identified as a significant source of heterogeneity, with cross-sectional, case-control, and cohort studies exhibiting varying effect sizes. This variation could be attributed to differences in study methodologies, sample sizes, and the timing of exposure assessment. Furthermore, there were geographical differences in the odds ratio across regions, with studies from Asia, Australia, and Africa showing the highest odds. These regional disparities may be influenced by variations in smoking prevalence, genetic susceptibility, cultural factors, and healthcare systems. These findings are consistent with the meta-analysis conducted by Huang et al. In their analysis, they also observed variations in the odds of ADHD among children in subgroup analyses based on certain study variables [21].

Although Egger’s test did not indicate significant publication bias, Begg’s test suggested some bias, which was addressed using the trim and fill method. This method estimated the effect size for potentially missing studies, resulting in a revised OR of 1.54 (1.40-1.70). This adjusted estimate reinforces the significant association between maternal tobacco smoking during pregnancy and ADHD in children, indicating that the observed effect is not solely due to publication bias. The sensitivity analysis further confirmed the robustness of the findings. This consistency underscores the reliability of the meta-analytic findings [49].

The findings of this meta-analysis have significant implications for public health and clinical practice. Given the association between maternal tobacco smoking during pregnancy and ADHD in children, there is a critical need for targeted tobacco smoking cessation programs for pregnant women [111,112]. Healthcare providers should emphasize the risks of tobacco smoking during prenatal visits and provide resources and support for smoking cessation [111]. Public health campaigns should also raise awareness about the potential long-term neurodevelopmental consequences of prenatal tobacco smoking. Policies aimed at reducing tobacco smoking prevalence among women of childbearing age could have a substantial impact on reducing the incidence of ADHD and other neurodevelopmental disorders.

Limitations

This meta-analysis, while strengthened by a large sample size and rigorous methodology, has limitations. The observational design of the included studies precludes causal inferences. Potential recall bias and underreporting may be present due to the reliance on self-reported maternal tobacco smoking during pregnancy. Further, the analysis did not differentiate between ADHD subtypes (predominantly inattentive, predominantly hyperactive-impulsive, and combined type), limiting the ability to identify subtype-specific risks associated with maternal tobacco smoking.

Future research

Future research should prioritize several key areas. Investigating the potential moderating roles of genetic and environmental factors on the relationship between prenatal tobacco exposure and ADHD is crucial. Additionally, exploring potential interactions between maternal smoking and other prenatal exposures (e.g., alcohol and drug use) would enhance our understanding of ADHD risk factors. Critically, examining whether the association between prenatal tobacco exposure and ADHD risk varies across ADHD subtypes is essential for a more nuanced understanding of the neurodevelopmental impact and could inform targeted interventions. Finally, further investigation into the underlying epigenetic mechanisms, such as DNA methylation alterations, is warranted. This could identify potential biomarkers and therapeutic targets for ADHD prevention and treatment.

Conclusion

This systematic review and meta-analysis demonstrate a significant association between maternal tobacco smoking during pregnancy and increased odds of ADHD in offspring. These findings underscore the importance of smoking cessation programs for pregnant women and broader public health interventions to reduce prenatal tobacco exposure. Future research should prioritize investigating causal mechanisms and potential interactions with other prenatal exposures. Reducing maternal smoking during pregnancy could substantially improve neurodevelopmental outcomes in children.

Supporting information

S1 File. Tables.

(DOCX)

S2 Checklist. PRISMA 2020 Checklist.

(DOCX)

Data Availability

All relevant data are within the paper and its Supporting Information files.

Funding Statement

The author(s) received no specific funding for this work.