Contemporary Epidemiology of Rising Atrial Septal Defect Trends Across USA 1991-2016 Dataset: A Combined Ecological Geotemporospatial and Causal Inferential Study.
Description
Background: Cardiovascular anomalies are the largest group of congenital anomalies and the major cause of death in young children, with a range of data linking rising atrial septal defect incidence (ASDI) with prenatal cannabis exposure. Objectives / Hypotheses. Is cannabis associated with ASDI in USA? Is this relationship causal? Methods: Geospatiotemporal cohort study, 1991-2016. Census populations of adults, babies, congenital anomalies, income and ethnicity. Drug exposure data on cigarettes, alcohol abuse, past month cannabis use, analgesia abuse and cocaine taken from National Survey of Drug Use and Health (78.9% response rate). Cannabinoid concentrations from Drug Enforcement Agency. Inverse probability weighted (ipw) regressions. Analysis conducted in R. Results. ASDI rose nationally three-fold from 27.4 to 82.8 / 10,000 births 1991-2014 during a period when tobacco and alcohol abuse were falling but cannabis was rising. States including Nevada, Kentucky, Mississippi and Tennessee had steeply rising epidemics (Time: Status β-estimate=10.72 (95%C.I. 8.39-13.05), P<2.0x10-16). ASDI was positively related to exposure to cannabis and most cannabinoids. Drug exposure data was near-complete from 2006 thus restricting spatial modelling from 2006-2014, N = 282. In geospatial regression models cannabis: alcohol abuse term was significant (β-estimate=19.44 (9.11, 29.77), P=2.2x10-4); no ethnic or income factors survived model reduction. Cannabis legalization was associated with a higher ASDI (Time: Status β-estimate=0.03 (0.01, 0.05), P=1.1x10-3). Weighted panel regression interactive terms including cannabis significant (from β-estimate=1418, (1080.6, 1755.4), P=7.3x10-15). Robust generalized linear models utilizing inverse probability weighting interactive terms including cannabis appear (from β-estimate=78.88, (64.38, 93.38), P=1.1x10-8). Marginal structural models with machine-aided SuperLearning association of ASDI with high v. low cannabis exposure R.R.=1.32 (1.28, 1.36). Model e-values mostly >1.5. Conclusions. ASDI is associated with cannabis use, frequency, intensity and legalization in a spatiotemporally significant manner, robust to socioeconomicodemographic adjustment and fulfilled causal criteria, consistent with multiple biological mechanisms and similar reports from Hawaii, Colorado, Canada and Australia. Not only are these results of concern in themselves, but they further imply that our list of the congenital teratology of cannabis is as yet incomplete, and highlight the particular cardiovascular toxicology of prenatal cannabinoid and drug exposure.