This study attempts to evaluate whether autism spectrum disorder in offspring is associated with maternal exposure to lithium in drinking water during pregnancy.
Key Points
Question
Is maternal prenatal exposure to lithium in drinking water associated with autism spectrum disorder (ASD) in offspring?
Findings
In this Danish nationwide population-based case-control study, the study team found that maternal exposure to higher levels of residential lithium in drinking water during pregnancy was associated with a moderate increase in ASD risk in the offspring. The findings remained robust after adjusting for several maternal neighborhood socioeconomic factors and air pollution exposures.
Meaning
This study suggests that naturally occurring lithium in drinking water may be a novel environmental risk factor for ASD development that requires further scrutiny.
Abstract
Importance
Lithium is a naturally occurring and trace element that has mood-stabilizing effects. Maternal therapeutic use of lithium has been associated with adverse birth outcomes. In animal models, lithium modulates Wnt/β-catenin signaling that is important for neurodevelopment. It is unknown whether exposure to lithium in drinking water affects brain health in early life.
Objective
To evaluate whether autism spectrum disorder (ASD) in offspring is associated with maternal exposure to lithium in drinking water during pregnancy.
Design, Setting, and Participants
This nationwide population-based case-control study in Denmark identified 8842 children diagnosed with ASD born from 2000 through 2013 and 43 864 control participants matched by birth year and sex from the Danish Medical Birth Registry. These data were analyzed from March 2021 through November 2022.
Exposures
Geocoded maternal residential addresses during pregnancy were linked to lithium level (range, 0.6 to 30.7 μg/L) in drinking water estimated using kriging interpolation based on 151 waterworks measurements of lithium across all regions in Denmark.
Main Outcomes and Measures
ASD diagnoses were ascertained using International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes recorded in the Danish Psychiatric Central Register. The study team estimated odds ratios (ORs) and 95% CIs for ASD according to estimated geocoded maternal exposure to natural source of lithium in drinking water as a continuous (per IQR) or a categorical (quartile) variable, adjusting for sociodemographic factors and ambient air pollutants levels. The study team also conducted stratified analyses by birth years, child’s sex, and urbanicity.
Results
A total of 8842 participants with ASD (male, 7009 [79.3%]) and 43 864 control participants (male, 34 749 [79.2%]) were studied. Every IQR increase in estimated geocoded maternal exposure to natural source of lithium in drinking water was associated with higher odds for ASD in offspring (OR, 1.23; 95% CI, 1.17-1.29). Elevated odds among offspring for ASD were estimated starting from the second quartile (7.36 to 12.67 μg/L) of estimated maternal exposure to drinking water with lithium and the OR for the highest quartile (more than 16.78 μg/L) compared with the reference group (less than 7.39 μg/L) was 1.46 (95% CI, 1.35-1.59). The associations were unchanged when adjusting for air pollution exposures and no differences were apparent in stratified analyses.
Conclusions and Relevance
Estimated maternal prenatal exposure to lithium from naturally occurring drinking water sources in Denmark was associated with an increased ASD risk in the offspring. This study suggests that naturally occurring lithium in drinking water may be a novel environmental risk factor for ASD development that requires further scrutiny.
Introduction
Lithium is a naturally occurring and trace element. Because of the mood-stabilizing effects of lithium,1,2 the compound has long been used as a psychiatric medication to treat bipolar disorders3,4 and depression.5,6 Exactly how lithium affects the brain in these psychiatric disorders remains unknown but multiple intersecting mechanisms have been suggested, such as lithium’s direct inhibition of glycogen synthase kinase-3β or effects on neurotrophic factors, neurotransmitters, oxidative metabolism, apoptosis, and second messenger systems.7,8,9,10,11,12,13,14 As a naturally occurring element, lithium is present in drinking water at low concentrations because of the weathering of minerals in the subsurface. Although food items also contain lithium,15 a major source of lithium intake in the general population is drinking water.16,17 Recent epidemiological studies suggested that chronic and low-dose ingestion of lithium from drinking water might have preventive effects on the incidence of dementia18,19,20 and suicide21,22,23 but not schizophrenia spectrum disorders and bipolar disorders.24
It is still debated whether lithium can be safely administered to treat bipolar disorders during pregnancy,25 as there is increasing evidence that lithium prescribed during pregnancy is associated with miscarriages26 and cardiac malformations in newborns.27 Since lithium can cross the placenta and the fetal blood-brain-barrier,28,29 it is biologically plausible that the compound influences fetal brain development, especially as lithium has been shown to modulate Wnt/β-catenin signaling, an important molecular pathway involved in neurodevelopment and autism.30 We conducted this study to estimate the association between prenatal exposure to lithium from drinking water and the risk for autism spectrum disorder (ASD) in Denmark using a population-based design.
Methods
This study was approved by the Danish Data Protection Agency and the institutional review board at University of California, Los Angeles. Informed consent was not required according to Danish law governing registry-based research studies with no participant contact. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines were followed.
Study Population
We conducted a population-based nested case-control study of ASD in Denmark31 with the source population defined as singleton births in Denmark born between 2000 and 2013 with a medical birth record available from the Danish Medical Birth Register.32 ASD cases were identified using the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (codes F84.0, F84.1, F84.5, F84.8, or F84.9) up to 2016 extracted from the Danish National Patient Register that contains information on admissions and outpatient/emergency department consultations from all general hospitals in Denmark33 and the Danish Psychiatric Central Register34 that holds records of admissions or outpatient visits to psychiatric hospitals in Denmark.35 Control participants were selected among singleton children born in Denmark during the same study period without having received a diagnosis for ASD and they were individually matched to cases on birth year and child’s sex at a ratio of 1:5. We also required that the children were born between 21 and 46 weeks of gestation with a birth weight of more than 500 g and less than 6800 g and that maternal addresses were recorded in the Danish Civil Registration System to estimate prenatal residential exposures to lithium from drinking water. A total of 8842 participants with ASD and 43 864 control participants were included in the statistical analyses (eFigure in the Supplement).
Lithium Exposure Assessment
We geocoded all maternal residential addresses (including dates of having moved) recorded in the Danish Civil Registration System during the interval of 9 months before until 9 months after the pregnancy.31 Each address, identified by municipality code, street code, and house number, was linked to the Danish Address Register to obtain geographical coordinates with a high level of precision (within 5 m).31 We successfully geocoded 99.6% of all addresses identified for all participants in this study.31
Groundwater is the source of all drinking water in Denmark, typically only with limited treatment and always without chlorination.36 We sampled the 144 largest public waterworks throughout Denmark in 2013 and analyzed for lithium levels using inductively couples mass spectrometry with a detection limit of 0.5 μg/L.17,37 Measurements from 7 additional waterworks recorded from 2009 to 2010 and analyzed by atomic absorption spectroscopy were also included to develop a kriging model. The mean lithium level was estimated to be 11.6 (SD, 6.8) μg/L, ranging from the lowest of 0.6 μg/L in Western Denmark to the highest of 30.7 μg/L in Eastern Denmark.21 Altogether, these 151 waterworks stand for the water supply of approximately half the Danish population.37 For 12 waterworks, we had data on lithium levels in the raw groundwater and the finished treated drinking water, indicating treatment does not alter lithium concentrations.37 A comparison with lithium levels in the same waterworks from 2016 to 2017 showed that concentrations were highly stable over time (R2 = 0.94 for repeated measurements), which is expected as the source of lithium is geogenic. From our waterworks measurements, we have also found that lithium measures are not correlated with zinc (R2 = 0.05) or nickel (R2 = 0.04) in our samples.
We used kriging to estimate the lithium levels for the entire country of Denmark, which has been described previously.18,21,38 The kriging map was used to compute mean lithium concentrations in each of the 275 Danish municipalities before the communal reform in 2007. For this study, geocoded maternal residential addresses for participants with ASD and control participants were linked to the kriging model estimates to compute participant-specific levels of average lithium in drinking water during pregnancy. Since lithium level estimates can be assumed to be stable over time within a municipality/grid cell, pregnant women’s residential drinking water levels are expected to be constant, unless they moved during pregnancy. To account for moving, we computed time-averaged lithium estimates for participants with multiple addresses (geocoded locations) during pregnancy.
Statistical Analysis
We estimated the odds ratios (ORs) and 95% CIs for offspring with ASD according to maternal residential lithium exposure in drinking water during pregnancy in unconditional logistic regression models adjusting for the matching variables (child’s sex and the exact birth year).39 Lithium level (in μg/L) was first analyzed continuously (per IQR increase) and we also generated estimates for exposure quartiles based on the distribution among control participants. We also used a restricted cubic spline model with 4 evenly distributed knots (at the 20th, 40th, 60th, and 80th percentiles) to investigate potential nonlinearity and thresholds for estimated effects. We adjusted for the following covariates in all analyses: maternal age (younger than 19, 19 to 25, 26 to 30, 31 to 35, older than 35 years), maternal smoking during pregnancy (yes or no), location of residence (Copenhagen, its suburbs, and large cities [Aarhus, Odense, Aalborg], provincial cities and rural towns/communities), neighborhood socioeconomic status measures, including the community employment status (ratio of unemployed to employed families based on the adult with the highest reported income in the year of birth), and a housing index (the percentage of council rented housing in the year of birth). In addition, we also adjusted for ambient air pollution levels for NO2 and PM2.5 during pregnancy, measures we previously generated for all study participants using the DEHM/UBM/AirGIS modeling system reflecting local pollution from street traffic in addition to contributions from urban and regional background pollution.40,41 Complete-case analyses were conducted because of the low number of missing values for all covariates used for confounder adjustment in our models.
In addition, we performed stratified analyses by child’s sex (male or female) to evaluate potential sex-differences.31 Moreover, we stratified the data by birth year (2000 to 2003, 2004 to 2007, 2007 to 2013). Since most drinking water measurements were conducted in 2013, if our assumption of time-constant lithium levels in birth years do not hold, we would expect a stronger nondifferential measurement error affecting results from the earlier birth years. We also stratified the data by the place of maternal residence at childbirth (Copenhagen and large cities vs provincial and rural towns/communities) to assess rural/urban differences. We performed tests of heterogeneity based on 2-tailed P values with a significance level of .05 for the product term of the continuous exposure variable and each of the binary potential effect modifiers in regression models. We also performed analyses that distinguished between subtypes of ASD (autistic disorder, Asperger syndrome, or pervasive developmental disorder-not otherwise specified). Additionally, we performed 2 quantitative bias analyses employing the E-value method42 and conducted a probabilistic bias adjustment43 to evaluate the potential influence from uncontrolled confounding bias. All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc).
Results
In Denmark, ASD is about 4 times more common in male than female children (79% vs 21%) (Table 1). The participants with ASD had a lower mean birth weight compared with control participants. The study team estimated a 23% increase in the odds for ASD in offspring (adjusted odds ratio [aOR], 1.23; 95% CI, 1.17-1.29) per IQR increase of the geospatial estimated lithium level in drinking water at maternal residence adjusting for maternal demographics, neighborhood socioeconomic factors, and ambient air pollutants (NO2 and PM2.5) (Table 2). Lithium exposures in the second and third quartiles (ranging from 7.4 to 16.8 μg/L) were associated with 24% to 26% higher odds for ASD diagnosis, and in the highest quartile (more than 16.8 μg/L) lithium was associated with 46% higher odds (aOR, 1.46; 95% CI, 1.35-1.59) compared with the lowest quartile as the reference (less than 7.4 μg/L). The spline model corroborated these results and suggested that ASD risk increased from the lower to higher exposure range with steeper slopes observed for the 5 to 10 μg/L and again the 15 to 25 μg/L ranges (Figure).
Table 1. Demographic and Perinatal Characteristics of Participants With Autism Spectrum Disorders (ASDs) and the Matched Population Control Participants, Denmark, 2000 to 2013.
| Characteristic | No. (%) | |
|---|---|---|
| Participants with ASD (n = 8842) | Control participants (n = 43 864)a | |
| Child’s sex | ||
| Male | 7009 (79.3) | 34 749 (79.2) |
| Female | 1833 (20.7) | 9115 (20.8) |
| Birth year | ||
| 2000-2003 | 4728 (53.5) | 23 407 (53.4) |
| 2004-2007 | 2943 (33.3) | 14 650 (33.4) |
| 2008-2013 | 1171 (13.2) | 5807 (13.2) |
| Mean (SD) | ||
| Birth weight, g | 3494.1 (655.2) | 3551.8 (578.7) |
| Gestational age, d | 275.8 (15.5) | 277.0 (13.1) |
| Maternal age, y | 29.9 (5.1) | 30.0 (4.8) |
| Location at birth | ||
| Copenhagen and large citiesb | 5004 (56.6) | 21 587 (49.2) |
| Provincial cities/towns | 3026 (34.2) | 16 980 (38.7) |
| Rural communities/small townsc | 812 (9.2) | 5289 (12.1) |
| Missing | 0 | 8 (.02) |
| Maternal smoking during pregnancy | ||
| No | 6562 (74.2) | 34 945 (79.7) |
| Yes | 2041 (23.1) | 7903 (18.0) |
| Missing | 239 (2.7) | 1016 (2.3) |
Control participants were matched to cases by child’s sex and birth year.
Includes Copenhagen and its suburbs, as well as large cities (Aarhus, Odense, and Aalborg).
Includes rural areas and towns with less than 10 000 inhabitants.
Table 2. Autism Spectrum Disorder (ASD) According to Lithium (Li) Levels in Maternal Residential Drinking Water During Pregnancy, Denmark, 2000 to 2013.
| Prenatal Li exposure | No. | OR (95% CI) | |||
|---|---|---|---|---|---|
| Participants with ASD | Control participants | Adjusteda | Adjustedb | Adjustedc | |
| Per IQR increase in Li (9.4 μg/L) | 8842 | 43 864 | 1.33 (1.28-1.39) | 1.24 (1.18-1.30) | 1.23 (1.17-1.29) |
| Li quartile, μg/L | |||||
| Q1 (≤7.4) | 1718 | 11 133 | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| Q2 (7.4-12.7) | 2185 | 10 931 | 1.30 (1.21-1.39) | 1.23 (1.15-1.32) | 1.26 (1.17-1.36) |
| Q3 (12.7-16.8) | 2089 | 10 471 | 1.29 (1.21-1.39) | 1.21 (1.12-1.31) | 1.24 (1.14-1.34) |
| Q4 (>16.8) | 2850 | 11 329 | 1.63 (1.53-1.74) | 1.48 (1.38-1.60) | 1.46 (1.35-1.59) |
Abbreviations: OR, odds ratio; Q, quarter.
Model adjusted for matching variables child’s sex and birth year.
Model adjusted for child’s sex and birth year, maternal age, location of birth, and neighborhood socioeconomic measures, including the community employment ratio and housing index and maternal smoking during pregnancy.
Model adjusted for child’s sex and birth year, maternal age, location of birth, and neighborhood socioeconomic measures, including the community employment ratio and housing index, maternal smoking during pregnancy, and air pollution (NO2, PM2.5) during pregnancy.
Figure. Odds Ratio for Autism Spectrum Disorder (ASD) in Childhood According to Continuous Lithium Levels in Maternal Residential Drinking Water During Pregnancy.
Odds ratio for ASD in childhood according to continuous lithium (Li, μg/L) levels in maternal residential drinking water during pregnancy estimated with a restricted cubic spline regression model. The reference level is set at the 25th percentile. The shading represents 95% CIs. Model adjusted for child’s sex, birth year, maternal age, location of birth, neighborhood community employment ratio and housing index, and maternal smoking during pregnancy.
In stratified analyses, the estimated effect size was slightly larger in female children and for the birth years 2000 to 2003, but the interaction P values were greater than .10 (Table 3). The associations between estimated maternal exposure to lithium in drinking water and childhood ASD were slightly stronger for mothers and children living in urban areas (aOR, 1.28; 95% CI, 1.18-1.38 for ASD per IQR increase in lithium levels), including Copenhagen and other larger cities, than for those residing in provincial towns/rural area (aOR, 1.14; 95% CI, 1.07-1.22 for ASD per IQR increase in lithium levels) (eTable 1 in the Supplement).
Table 3. Autism Spectrum Disorder (ASD) According to Lithium (Li) Levels in Maternal Residential Drinking Water During Pregnancy in Denmark, Stratified by Child’s Sex and Birth Year.
| Prenatal Li, μg/La | No. | OR (95% CI) | ||
|---|---|---|---|---|
| Participants with ASD | Control participants | Adjustedb | Adjustedc | |
| Stratified by child’s sex | ||||
| Male | ||||
| Per IQR increase (9.4 μg/L) | 7009 | 34 749 | 1.21 (1.15-1.28) | 1.20 (1.14-1.27) |
| Q1 (≤7.4) | 1376 | 8794 | 1 [Reference] | 1 [Reference] |
| Q2 (7.4-12.7) | 1758 | 8634 | 1.25 (1.15-1.35) | 1.28 (1.18-1.39) |
| Q3 (12.7-16.8) | 1646 | 8334 | 1.19 (1.09-1.30) | 1.22 (1.11-1.33) |
| Q4 (>16.8) | 2229 | 8987 | 1.46 (1.34-1.59) | 1.45 (1.32-1.58) |
| Female | ||||
| Per IQR increase (9.4 μg/L) | 1833 | 9115 | 1.34 (1.21-1.49) | 1.32 (1.18-1.47) |
| Q1 (≤7.4) | 342 | 2339 | 1 [Reference] | 1 [Reference] |
| Q2 (7.4-12.7) | 427 | 2297 | 1.16 (0.99-1.36) | 1.20 (1.01-1.41) |
| Q3 (12.7-16.8) | 443 | 2137 | 1.31 (1.10-1.55) | 1.32 (1.11-1.58) |
| Q4 (>16.8) | 621 | 2342 | 1.58 (1.34-1.87) | 1.53 (1.28-1.82) |
| Stratified by child’s birth year | ||||
| Years 2000-2003 | ||||
| Per IQR increase (9.4 μg/L) | 4728 | 23 407 | 1.30 (1.22-1.39) | 1.29 (1.20-1.38) |
| Q1 (≤7.4) | 934 | 6017 | 1 [Reference] | 1 [Reference] |
| Q2 (7.4-12.7) | 1240 | 5930 | 1.31 (1.19-1.44) | 1.33 (1.21-1.47) |
| Q3 (12.7-16.8) | 1022 | 5540 | 1.16 (1.04-1.29) | 1.18 (1.05-1.31) |
| Q4 (>16.8) | 1532 | 5920 | 1.59 (1.43-1.76) | 1.57 (1.41-1.75) |
| Years 2004-2007 | ||||
| Per IQR increase (9.4 μg/L) | 2943 | 14 650 | 1.19 (1.09-1.29) | 1.16 (1.07-1.27) |
| Q1 (≤7.4) | 568 | 3660 | 1 [Reference] | 1 [Reference] |
| Q2 (7.4-12.7) | 703 | 3643 | 1.17 (1.03-1.32) | 1.18 (1.03-1.34) |
| Q3 (12.7-16.8) | 728 | 3523 | 1.22 (1.07-1.39) | 1.21 (1.06-1.39) |
| Q4 (>16.8) | 944 | 3824 | 1.41 (1.24-1.60) | 1.37 (1.20-1.57) |
| Years 2008-2013 | ||||
| Per IQR increase (9.4 μg/L) | 1171 | 5807 | 1.15 (1.00-1.31) | 1.12 (0.97-1.29) |
| Q1 (≤7.4) | 216 | 1456 | 1 [Reference] | 1 [Reference] |
| Q2 (7.4-12.7) | 242 | 1358 | 1.08 (0.87-1.33) | 1.12 (0.90-1.39) |
| Q3 (12.7-16.8) | 339 | 1408 | 1.41 (1.14-1.75) | 1.45 (1.16-1.81) |
| Q4 (>16.8) | 374 | 1585 | 1.30 (1.04-1.61) | 1.25 (0.99-1.57) |
Abbreviations: OR, odds ratio; Q, quarter.
Lithium quartiles: Q1, less than or equal to 7.4; Q2, 7.4 to 12.7; Q3, 12.7 to 16.8; Q4, more than 16.8 μg/L.
Model adjusted for child’s sex (in birth year stratified analyses), the exact birth year (in both sex or birth-year categories stratified analyses), maternal age, location of birth, neighborhood community employment ratio and housing index, and maternal smoking during pregnancy.
Additionally adjusted for ambient air pollution (NO2, PM2.5) during pregnancy based on model b.
Analyses by ASD subtypes showed similar associations for autistic disorder cases (aOR, 1.24; 95% CI, 1.15-1.34 per IQR increase of estimated maternal exposure to lithium levels and aOR, 1.49; 95% CI, 1.32-1.68 comparing the highest vs the lowest quartile of lithium levels), while for Asperger syndrome cases, the associations seemed to be driven predominantly by the highest exposure quartile. The effect sizes were smaller for pervasive developmental disorder-not otherwise specified, but the effect estimates were still elevated (aOR, 1.09; 95% CI, 1.00-1.18 per IQR increase of lithium levels and aOR, 1.23; 95% CI, 1.07-1.41 comparing the highest vs the lowest quartile of lithium levels) (Table 4).
Table 4. Autism Spectrum Disorders (ASDs) According to Lithium (Li) Levels in Maternal Residential Drinking Water During Pregnancy in Denmark, by ASD Subtypes.
| Prenatal Li, μg/La | No. | OR (95% CI) | ||
|---|---|---|---|---|
| Participants with ASD | Control participants | Adjustedb | Adjustedc | |
| Autistic disorder | ||||
| Per IQR increase (9.4 μg/L) | 3962 | 19 240 | 1.26 (1.18-1.36) | 1.24 (1.15-1.34) |
| Q1 (≤7.4) | 750 | 4833 | 1 [Reference] | 1 [Reference] |
| Q2 (7.4-12.7) | 969 | 4683 | 1.31 (1.18-1.46) | 1.33 (1.19-1.48) |
| Q3 (12.7 16.8) | 951 | 4549 | 1.34 (1.19-1.50) | 1.34 (1.19-1.51) |
| Q4 (>16.8) | 1292 | 5175 | 1.53 (1.36-1.71) | 1.49 (1.32-1.68) |
| Asperger syndrome | ||||
| Per IQR increase (9.4 μg/L) | 1580 | 7660 | 1.38 (1.23-1.55) | 1.34 (1.19-1.51) |
| Q1 (≤7.4) | 311 | 1968 | 1 [Reference] | 1 [Reference] |
| Q2 (7.4-12.7) | 358 | 1966 | 1.06 (0.90-1.26) | 1.09 (0.91-1.30) |
| Q3 (12.7 16.8) | 300 | 1843 | 0.90 (0.74-1.09) | 0.90 (0.74-1.10) |
| Q4 (>16.8) | 611 | 1883 | 1.73 (1.45-2.06) | 1.63 (1.35-1.97) |
| Pervasive developmental disorder-not otherwise specified | ||||
| Per IQR increase (9.4 μg/L) | 3075 | 14 937 | 1.08 (0.99-1.17) | 1.09 (1.00-1.18) |
| Q1 (≤7.4) | 650 | 3761 | 1 [Reference] | 1 [Reference] |
| Q2 (7.4-12.7) | 767 | 3754 | 1.11 (0.99-1.25) | 1.16 (1.03-1.31) |
| Q3 (12.7 16.8) | 774 | 3617 | 1.16 (1.02-1.32) | 1.21 (1.06-1.39) |
| Q4 (>16.8) | 884 | 3805 | 1.21 (1.07-1.38) | 1.23 (1.07-1.41) |
Abbreviation: OR, odd ratio; Q, quarter.
Lithium quartiles: Q1, less than or equal to 7.4; Q2, 7.4 to 12.7; Q3, 12.7 to 16.8; Q4, more than 16.8 μg/L.
Model adjusted for child’s sex, birth year, maternal age, location of birth, neighborhood community employment ratio and housing index, and maternal smoking during pregnancy.
Additionally adjusted for ambient air pollution (NO2, PM2.5) during pregnancy based on model b.
The E-value was 2.02 for the observed association between ASD and the highest quartile of lithium exposure compared with the lowest (OR, 1.46; 95% CI, 1.35-1.59). Using probabilistic bias adjustments, the study team presents a range of effect estimates after accounting for a presumed unmeasured confounder that is positively associated with the exposure group and the outcome (eTable 2 in the Supplement). The adjusted effect estimate for lithium becomes null only when the unmeasured confounder is strongly associated with ASD (eg, OR, 5) and nearly 2 times more common in the exposed than the unexposed group.
Discussion
In this population-based study of ASD in Denmark, we found that estimated maternal exposure to higher levels of residential lithium in drinking water during pregnancy was associated with a moderate increase in ASD risk in the offspring while adjusting for several sociodemographic and neighborhood factors, including air pollution. Although we cannot rule out confounding due to unmeasured factors in our analyses, a single uncontrolled confounding factor is unlikely to explain away our findings unless it is strongly associated with ASD and lithium exposure. Overall, our study suggests that naturally occurring lithium in drinking water may be a novel environmental risk factor for ASD development that requires further scrutiny.
Research that examines health effects associated with low-dose exposure to lithium from drinking water sources has received increased attention,16,44,45 including recent studies of adults’ mental health conducted in Denmark that were based on the same waterworks lithium measurements employed in this study.18,21,38 Concerns about exposure to therapeutic doses of lithium in pregnancy, including increased risks for miscarriage26 and cardiac malformations in newborns, have been raised.27 Only 1 prior study of 194 mother-child pairs recruited from the northern Argentinean Andes, a region with varying and high drinking water lithium concentrations (5 to 1600 μg/L), has focused on studying the effects of lithium from drinking water on fetal growth.46 This study found increasing maternal blood and urinary lithium levels to be inversely associated with ultrasound-estimated fetal size (body, head, and femur) in the second trimester and also a decrease in birth length.46 The study also reported that lithium levels detected in the drinking water correlated moderately well with the lithium biomarkers in whole blood and urine (r = 0.40-0.44; P < .001). Our study found the higher levels of prenatal lithium exposure in Denmark to be associated with childhood ASD risk, even though the exposure range and maximum of lithium detected in Danish drinking water supplies (0.6 to 30.7 μg/L) were much lower than in Argentina. These observations call for additional epidemiological studies to examine lithium-exposure related ASD risk, as well as adverse fetal development, including studies that address dose response and gestational timing of exposures.
Strengths and Limitations
Strengths of our study are that it is a large nationwide, population-based study of ASD and that the cases were identified from national registries.31 This record-linkage study avoids any self-selection or recall bias. ASD diagnoses made in the Danish Psychiatric Central Registry are found to have high validity for epidemiologic research.47 Complete maternal addresses were ascertained from the Danish Civil Registration System allowing residential environmental exposure assessments during pregnancy, a critical and relevant time window for ASD development.31 Bottled water consumption is low in Denmark (ranks as the second lowest in Europe),48 making it an ideal location to study potential health effect of lithium exposure from drinking tap water. The residential lithium level in drinking water is modeled based on lithium measurements of 151 waterworks from 2009 to 2010 and 2013, spatially covering all regions of Denmark.18,21 We were able to adjust for several potential social-neighborhood confounding factors and also for ambient air pollution levels at maternal residential locations.31
Our study also has several limitations. First, this registry study lacks information on dietary and lifestyle factors, including habits of water consumption. This study did not measure the actual sources of water the mothers consumed and did not consider childhood exposure. However, there are no established lifestyle risk factors for autism, thus these are unlikely to confound the associations with environmental sources of lithium, of which study participants were generally unaware.49 Potential confounders may include elements and natural compounds of drinking water sources that are geospatially correlated with lithium, especially if they are of the same geogenic origin as lithium. We have checked that lithium does not co-occur with zinc and nickel in our waterworks measurements. However, lithium may geographically correlate with iodine in groundwater (relatively higher in the eastern part of Denmark). Previously, a principle component analysis in a nationwide groundwater study has also revealed that elevated iodine levels were associated with elevated lithium.50 Nevertheless, low-level intake of iodine may lead to thyroid deficiency and adversely affect neurodevelopment;51,52 thus, residual confounding by iodine could potentially bias the effect estimates toward the null. Moreover, arsenic in some waterworks detected at above national and guideline values53 could potentially induce adverse health effects, but these higher arsenic levels are mostly detected for rural communities and small towns in Denmark.54 We also do not have data on maternal use of lithium medication, but the prevalence of such use is very low (less than 0.1%) and likely not associated with drinking water lithium content.55 Based on our quantitative bias analysis, each of the potential confounding factors discussed are unlikely to be strongly enough associated with ASD and with lithium exposure such that this would nullify the observed associations. We focused on lithium from residential drinking water and not from other exposure sources, such as intake of water at workplaces or via diets. Thus, the cumulative effect of lithium exposure from multiple sources cannot be assessed in this study. Additionally, our results may not be transportable to populations exposed to higher lithium concentrations outside the range of our data.
Conclusions
In summary, we found that geospatial model-based and estimated maternal pregnancy exposure to lithium from naturally occurring drinking water sources in Denmark was associated with ASD risk in the offspring. This study suggests that naturally occurring lithium in drinking water may be a novel environmental risk factor for ASD development that requires further scrutiny.
eFigure. Flowchart for the autism spectrum disorders (ASD) cases and controls selection
eTable 1. Odds ratio (OR) and 95% confidence interval (CI) for autism spectrum disorders (ASD) according to lithium (Li) levels in maternal residential drinking water during pregnancy in Denmark, stratified by the location of birth
eTable 2. The adjusted odds ratio (OR) comparing the highest vs. the lowest lithium (Li) exposure quartile and autism spectrum disorders (ASD), accounting for an unmeasured confounder (the observed OR = 1.46, 95% CI 1.35, 1.59)
Data sharing statement
References
- 1.Platman SR, Fieve RR. Biochemical aspects of lithium in affective disorders. Arch Gen Psychiatry. 1968;19(6):659-663. doi: 10.1001/archpsyc.1968.01740120019003 [DOI] [PubMed] [Google Scholar]
- 2.Preskorn SH, Irwin GH, Simpson S, Friesen D, Rinne J, Jerkovich G. Medical therapies for mood disorders alter the blood-brain barrier. Science. 1981;213(4506):469-471. doi: 10.1126/science.7244645 [DOI] [PubMed] [Google Scholar]
- 3.Goodwin FK, Fireman B, Simon GE, Hunkeler EM, Lee J, Revicki D. Suicide risk in bipolar disorder during treatment with lithium and divalproex. JAMA. 2003;290(11):1467-1473. doi: 10.1001/jama.290.11.1467 [DOI] [PubMed] [Google Scholar]
- 4.Gelenberg AJ, Kane JM, Keller MB, et al. Comparison of standard and low serum levels of lithium for maintenance treatment of bipolar disorder. N Engl J Med. 1989;321(22):1489-1493. doi: 10.1056/NEJM198911303212201 [DOI] [PubMed] [Google Scholar]
- 5.Mendels J. Lithium in the treatment of depression. Am J Psychiatry. 1976;133(4):373-378. doi: 10.1176/ajp.133.4.373 [DOI] [PubMed] [Google Scholar]
- 6.Swann AC, Bowden CL, Morris D, et al. Depression during mania. treatment response to lithium or divalproex. Arch Gen Psychiatry. 1997;54(1):37-42. doi: 10.1001/archpsyc.1997.01830130041008 [DOI] [PubMed] [Google Scholar]
- 7.Rowse AL, Naves R, Cashman KS, et al. Lithium controls central nervous system autoimmunity through modulation of IFN-γ signaling. PLoS One. 2012;7(12):e52658. doi: 10.1371/journal.pone.0052658 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Beaulieu JM, Sotnikova TD, Yao WD, et al. Lithium antagonizes dopamine-dependent behaviors mediated by an AKT/glycogen synthase kinase 3 signaling cascade. Proc Natl Acad Sci U S A. 2004;101(14):5099-5104. doi: 10.1073/pnas.0307921101 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Pert A, Rosenblatt JE, Sivit C, Pert CB, Bunney WE Jr. Long-term treatment with lithium prevents the development of dopamine receptor supersensitivity. Science. 1978;201(4351):171-173. doi: 10.1126/science.566468 [DOI] [PubMed] [Google Scholar]
- 10.Gallager DW, Pert A, Bunney WE Jr. Haloperidol-induced presynaptic dopamine supersensitivity is blocked by chronic lithium. Nature. 1978;273(5660):309-312. doi: 10.1038/273309a0 [DOI] [PubMed] [Google Scholar]
- 11.Morris G, Berk M. The putative use of lithium in Alzheimer’s disease. Curr Alzheimer Res. 2016;13(8):853-861. doi: 10.2174/1567205013666160219113112 [DOI] [PubMed] [Google Scholar]
- 12.Manji HK, Moore GJ, Chen G. Clinical and preclinical evidence for the neurotrophic effects of mood stabilizers: implications for the pathophysiology and treatment of manic-depressive illness. Biol Psychiatry. 2000;48(8):740-754. doi: 10.1016/S0006-3223(00)00979-3 [DOI] [PubMed] [Google Scholar]
- 13.Phiel CJ, Wilson CA, Lee VM, Klein PS. GSK-3alpha regulates production of Alzheimer’s disease amyloid-beta peptides. Nature. 2003;423(6938):435-439. doi: 10.1038/nature01640 [DOI] [PubMed] [Google Scholar]
- 14.Caccamo A, Oddo S, Tran LX, LaFerla FM. Lithium reduces tau phosphorylation but not a beta or working memory deficits in a transgenic model with both plaques and tangles. Am J Pathol. 2007;170(5):1669-1675. doi: 10.2353/ajpath.2007.061178 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Enderle J, Klink U, di Giuseppe R, et al. Plasma lithium levels in a general population: a cross-sectional analysis of metabolic and dietary correlates. Nutrients. 2020;12(8):2489. doi: 10.3390/nu12082489 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Eyre-Watt B, Mahendran E, Suetani S, Firth J, Kisely S, Siskind D. The association between lithium in drinking water and neuropsychiatric outcomes: a systematic review and meta-analysis from across 2678 regions containing 113 million people. Aust N Z J Psychiatry. 2021;55(2):139-152. doi: 10.1177/0004867420963740 [DOI] [PubMed] [Google Scholar]
- 17.Voutchkova DD, Hansen B, Ernstsen V, Kristiansen SM. Nationwide drinking water sampling campaign for exposure assessments in Denmark. Int J Environ Res Public Health. 2018;15(3):467. doi: 10.3390/ijerph15030467 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kessing LV, Gerds TA, Knudsen NN, et al. Association of lithium in drinking water with the incidence of dementia. JAMA Psychiatry. 2017;74(10):1005-1010. doi: 10.1001/jamapsychiatry.2017.2362 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Fajardo VA, Fajardo VA, LeBlanc PJ, MacPherson REK. Examining the relationship between trace lithium in drinking water and the rising rates of age-adjusted Alzheimer’s disease mortality in Texas. J Alzheimers Dis. 2018;61(1):425-434. doi: 10.3233/JAD-170744 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Baranowski BJ, Hayward GC, Fajardo VA, MacPherson REK. Increased prevalence of obesity/type 2 diabetes and lower levels of lithium in rural Texas counties may explain greater Alzheimer’s disease risk. J Alzheimers Dis. 2018;64(1):303-308. doi: 10.3233/JAD-171150 [DOI] [PubMed] [Google Scholar]
- 21.Knudsen NN, Schullehner J, Hansen B, et al. Lithium in drinking water and incidence of suicide: a nationwide individual-level cohort study with 22 years of follow-up. Int J Environ Res Public Health. 2017;14(6):627. doi: 10.3390/ijerph14060627 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Ohgami H, Terao T, Shiotsuki I, Ishii N, Iwata N. Lithium levels in drinking water and risk of suicide. Br J Psychiatry. 2009;194(5):464-465. doi: 10.1192/bjp.bp.108.055798 [DOI] [PubMed] [Google Scholar]
- 23.Vita A, De Peri L, Sacchetti E. Lithium in drinking water and suicide prevention: a review of the evidence. Int Clin Psychopharmacol. 2015;30(1):1-5. doi: 10.1097/YIC.0000000000000048 [DOI] [PubMed] [Google Scholar]
- 24.Schullehner J, Paksarian D, Hansen B, et al. Lithium in drinking water associated with adverse mental health effects. Schizophr Res. 2019;210:313-315. doi: 10.1016/j.schres.2019.06.016 [DOI] [PubMed] [Google Scholar]
- 25.Fornaro M, Maritan E, Ferranti R, et al. Lithium exposure during pregnancy and the postpartum period: a systematic review and meta-analysis of safety and efficacy outcomes. Am J Psychiatry. 2020;177(1):76-92. doi: 10.1176/appi.ajp.2019.19030228 [DOI] [PubMed] [Google Scholar]
- 26.Diav-Citrin O, Shechtman S, Tahover E, et al. Pregnancy outcome following in utero exposure to lithium: a prospective, comparative, observational study. Am J Psychiatry. 2014;171(7):785-794. doi: 10.1176/appi.ajp.2014.12111402 [DOI] [PubMed] [Google Scholar]
- 27.Patorno E, Huybrechts KF, Bateman BT, et al. Lithium use in pregnancy and the risk of cardiac malformations. N Engl J Med. 2017;376(23):2245-2254. doi: 10.1056/NEJMoa1612222 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Harari F, Ronco AM, Concha G, et al. Early-life exposure to lithium and boron from drinking water. Reprod Toxicol. 2012;34(4):552-560. doi: 10.1016/j.reprotox.2012.08.009 [DOI] [PubMed] [Google Scholar]
- 29.Forester BP, Streeter CC, Berlow YA, et al. Brain lithium levels and effects on cognition and mood in geriatric bipolar disorder: a lithium-7 magnetic resonance spectroscopy study. Am J Geriatr Psychiatry. 2009;17(1):13-23. doi: 10.1097/JGP.0b013e318172b3d0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Mulligan KA, Cheyette BN. Neurodevelopmental perspectives on wnt signaling in psychiatry. Mol Neuropsychiatry. 2017;2(4):219-246. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Ritz B, Liew Z, Yan Q, et al. Air pollution and autism in Denmark. Environ Epidemiol. 2018;2(4):e028. doi: 10.1097/EE9.0000000000000028 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Bliddal M, Broe A, Pottegård A, Olsen J, Langhoff-Roos J. The Danish Medical Birth Register. Eur J Epidemiol. 2018;33(1):27-36. doi: 10.1007/s10654-018-0356-1 [DOI] [PubMed] [Google Scholar]
- 33.Schmidt M, Schmidt SA, Sandegaard JL, Ehrenstein V, Pedersen L, Sørensen HT. The Danish National Patient Registry: a review of content, data quality, and research potential. Clin Epidemiol. 2015;7:449-490. doi: 10.2147/CLEP.S91125 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Pedersen CB. The Danish Civil Registration System. Scand J Public Health. 2011;39(7)(suppl):22-25. doi: 10.1177/1403494810387965 [DOI] [PubMed] [Google Scholar]
- 35.Mors O, Perto GP, Mortensen PB. The Danish Psychiatric Central Research Register. Scand J Public Health. 2011;39(7)(suppl):54-57. doi: 10.1177/1403494810395825 [DOI] [PubMed] [Google Scholar]
- 36.Schullehner J. Danish water supply areas and their links to water production facilities: an open-access data set. Accessed February 24, 2023. https://dataverse.geus.dk/dataset.xhtml?persistentId=doi:10.22008/FK2/I5R1SS
- 37.Voutchkova DDSJ, Knudsen NN, Jørgensen LF, Ersbøll AK, Kristiansen SM, Hansen B. Exposure to selected geogenic trace elements (I, Li, and Sr) from drinking water in Denmark. Geosciences (Basel). 2015;5(1):45-66. doi: 10.3390/geosciences5010045 [DOI] [Google Scholar]
- 38.Kessing LV, Gerds TA, Knudsen NN, et al. Lithium in drinking water and the incidence of bipolar disorder: a nation-wide population-based study. Bipolar Disord. 2017;19(7):563-567. doi: 10.1111/bdi.12524 [DOI] [PubMed] [Google Scholar]
- 39.Pearce N. Analysis of matched case-control studies. BMJ. 2016;352:i969. doi: 10.1136/bmj.i969 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Jensen S, Ketzel M, Becker T, et al. High resolution multi-scale air quality modelling for all streets in Denmark. Transp Res Part D Transp Environ. 2017;52:322-339. doi: 10.1016/j.trd.2017.02.019 [DOI] [Google Scholar]
- 41.Khan J, Kakosimos K, Raaschou-Nielsen O, et al. Development and performance evaluation of new AirGIS—a GIS based air pollution and human exposure modelling system. Atmos Environ. 2019;198:102-121. doi: 10.1016/j.atmosenv.2018.10.036 [DOI] [Google Scholar]
- 42.VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the e-value. Ann Intern Med. 2017;167(4):268-274. doi: 10.7326/M16-2607 [DOI] [PubMed] [Google Scholar]
- 43.Arah OA, Chiba Y, Greenland S. Bias formulas for external adjustment and sensitivity analysis of unmeasured confounders. Ann Epidemiol. 2008;18(8):637-646. doi: 10.1016/j.annepidem.2008.04.003 [DOI] [PubMed] [Google Scholar]
- 44.Shimodera S, Koike S, Ando S, et al. Lithium levels in tap water and psychotic experiences in a general population of adolescents. Schizophr Res. 2018;201:294-298. doi: 10.1016/j.schres.2018.05.019 [DOI] [PubMed] [Google Scholar]
- 45.Brown EE, Gerretsen P, Pollock B, Graff-Guerrero A. Psychiatric benefits of lithium in water supplies may be due to protection from the neurotoxicity of lead exposure. Med Hypotheses. 2018;115:94-102. doi: 10.1016/j.mehy.2018.04.005 [DOI] [PubMed] [Google Scholar]
- 46.Harari F, Langeén M, Casimiro E, et al. Environmental exposure to lithium during pregnancy and fetal size: a longitudinal study in the Argentinean Andes. Environ Int. 2015;77:48-54. doi: 10.1016/j.envint.2015.01.011 [DOI] [PubMed] [Google Scholar]
- 47.Lauritsen MB, Jørgensen M, Madsen KM, et al. Validity of childhood autism in the Danish Psychiatric Central Register: findings from a cohort sample born 1990-1999. J Autism Dev Disord. 2010;40(2):139-148. doi: 10.1007/s10803-009-0818-0 [DOI] [PubMed] [Google Scholar]
- 48.Rygaard M, Arvin E, Binning PJ. The valuation of water quality: effects of mixing different drinking water qualities. Water Res. 2009;43(5):1207-1218. doi: 10.1016/j.watres.2008.12.014 [DOI] [PubMed] [Google Scholar]
- 49.Weisskopf MG, Webster TF. Trade-offs of personal versus more proxy exposure measures in environmental epidemiology. Epidemiology. 2017;28(5):635-643. doi: 10.1097/EDE.0000000000000686 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Voutchkova DD, Kristiansen SM, Hansen B, Ernstsen V, Sørensen BL, Esbensen KH. Iodine concentrations in Danish groundwater: historical data assessment 1933-2011. Environ Geochem Health. 2014;36(6):1151-1164. doi: 10.1007/s10653-014-9625-4 [DOI] [PubMed] [Google Scholar]
- 51.Abel MH, Caspersen IH, Meltzer HM, et al. Suboptimal maternal iodine intake is associated with impaired child neurodevelopment at 3 years of age in the Norwegian mother and child cohort study. J Nutr. 2017;147(7):1314-1324. doi: 10.3945/jn.117.250456 [DOI] [PubMed] [Google Scholar]
- 52.Dineva M, Fishpool H, Rayman MP, Mendis J, Bath SC. Systematic review and meta-analysis of the effects of iodine supplementation on thyroid function and child neurodevelopment in mildly-to-moderately iodine-deficient pregnant women. Am J Clin Nutr. 2020;112(2):389-412. doi: 10.1093/ajcn/nqaa071 [DOI] [PubMed] [Google Scholar]
- 53.Ramsay L, Petersen MM, Hansen B, et al. Drinking water criteria for arsenic in high-income, low-dose countries: the effect of legislation on public health. Environ Sci Technol. 2021;55(6):3483-3493. doi: 10.1021/acs.est.0c03974 [DOI] [PubMed] [Google Scholar]
- 54.Richter F, Kloster S, Wodschow K, et al. Maternal exposure to arsenic in drinking water and risk of congenital heart disease in the offspring. Environ Int. 2022;160:107051. doi: 10.1016/j.envint.2021.107051 [DOI] [PubMed] [Google Scholar]
- 55.Munk-Olsen T, Liu X, Viktorin A, et al. Maternal and infant outcomes associated with lithium use in pregnancy: an international collaborative meta-analysis of six cohort studies. Lancet Psychiatry. 2018;5(8):644-652. doi: 10.1016/S2215-0366(18)30180-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eFigure. Flowchart for the autism spectrum disorders (ASD) cases and controls selection
eTable 1. Odds ratio (OR) and 95% confidence interval (CI) for autism spectrum disorders (ASD) according to lithium (Li) levels in maternal residential drinking water during pregnancy in Denmark, stratified by the location of birth
eTable 2. The adjusted odds ratio (OR) comparing the highest vs. the lowest lithium (Li) exposure quartile and autism spectrum disorders (ASD), accounting for an unmeasured confounder (the observed OR = 1.46, 95% CI 1.35, 1.59)
Data sharing statement