America Addresses Two Epidemics: Cannabis and Coronavirus and their Interactions: An Ecological Geospatial Study

Albert S Reece; Gary K Hulse

Given that many terms in the five domains of Supplementary Table 3 Table 2 presents final interactive spatial models after reduction via the spml, spgm and spreml algorithms. Interestingly travel, cannabis and opioid pain relievers are found to be significant in all final models. Terms including cannabis are most significant in the spml model, from P=7.3x10 -15 . Cannabis alone is significant (P=0.0365) in the spgm model, a technique which is sensitive to short panel datasets of this type. Cannabis is included in seven of nine terms, eight of twelve terms and seven of nine terms in the three models respectively. Interactions between last month cannabis use and daily cannabis use are included in three terms in each model. Tobacco, binge alcohol, cocaine and amphetamines did not appear in any final spatial models. Study of spreml model error structure confirmed that the full error structure (sem2srre without lagging) was indeed appropriate.