Most of the imputation strategies resulted in a significant improvement in estimates of substance use data when compared to estimates relying on complete case data, which was especially true in the case of a ‘rare’ outcome such as heroin use. However, MI performed the most favorably regardless of the prevalence of the outcome or the level of missingness. Missing data resulting from combining data across cohorts more plausibly resembles an MAR data mechanism given that missing substance use data is related to other observed data, namely cohort characteristics such as age and HIV status. We used plausible proxy cohort characteristics such as age, employment status, and other substance use variables to inform the imputation process, which likely results in both gains in effciency and reductions in bias. Te finding that five imputation data sets performed nearly as well as multiple imputation with twenty data sets addresses the practical decision of how many imputations are needed. While some of the original work in this area suggests that between two and ten imputations are sufficient, others have suggested that more than ten and up to 100 imputations may be needed if the fraction of missing information is large. Our findings suggest that efficiency can be maintained with smaller imputed data sets, even under circumstances where up to 50% of the data are being imputed. We recognize that if MI works in all scenarios, then having to make a decision to use another imputation strategy may be unnecessary. However, because the goal is to implement these strategies in the context of cross protocol analyses requiring harmonization across multiple cohorts with thousands of observations this can create a number of practical issues. First, researchers conducting cross-cohort analyses will need to have the technical expertise needed to analyze MI data sets which requires running statistical models on each of the imputed data sets and then pooling the resulting model parameter estimates. So, assuming twenty imputations are done, then this step will result in running statistical models on twenty sets of data, obtaining the twenty resulting parameter estimates,greenhouse rolling racks which then are pooled into a single pooled estimate.
Furthermore, practical considerations also come in to play given the potentially large sample sizes along with imputations creating up to twenty data sets. Te potential impact on computer processing power or statistical programming requirements may make such analyses difficult for many investigators. Te findings of this study should be interpreted in light of some of the limitations. We utilized a cross sectional view of the data and did not give consideration to a monotone missing data pattern resulting from loss to follow-up given that imputation for missing data resulting from loss to follow-up was not the goal of this study. Assessment of substance use was based on self-report and any response bias introduced as part of the data collection are not corrected for as part of these analyses. However, our use of computer assisted self-interview may have helped to improve the validity of the self-reported information. Nonetheless, our imputations were strengthened by the use of auxiliary variables. In particular our use of other substance use data is particularly relevant in that it will allow us to use participants’ reported substance use patterns to inform our imputations. A restrictive strategy that does not make use of auxiliary variables is less informative and as others have noted the more inclusive strategy reduces the chance of inadvertent omission of important causes of missing data with resulting gains in efficiency and reduction in bias. Te only cost here is the availability of data for the selected auxiliary variables.Tobacco smoking results in millions of preventable deaths each year worldwide. Nicotine, the main psychoactive component in tobacco, is considered to be responsible for the development and maintenance of dependence in humans. Nicotine’s effects on adolescent development have become of increasing concern given the emergence of e-cigarettes, which deliver vaporized nicotine . According to a nationwide CDC survey, ~30–45% of high school students self-reported prior use of cigarettes, vaporized nicotine products, and/or cannabis.
Given that legalization of recreational cannabis across states since the time of this survey, the number of adolescents exposed to this drug will likely continue to increase through both primary and second-hand exposure. Importantly, studies in humans examining co-use of these drugs have found that individuals who reported smoking both cannabis and tobacco cigarettes consumed more cigarettes than those using tobacco alone. Furthermore, the practice of mulling has been reported as frequently occurring in adolescent users, with high incidence among daily cigarette smokers in some populations. Interestingly, chronic male cannabis users show decreased activation of the caudate nucleus in relation to reward anticipation as compared to nicotine users and nonsmokers, suggesting altered function of reward-related circuitries dependent on prior drug exposure. Chronic use of cannabis during adolescence has also been linked to an elevated risk of psychosis, anxiety disorders, and depression. For instance, Crane and colleagues found that symptoms of depression were positively correlated with both cannabis use and tobacco smoking frequency in male, but not female, subjects. In contrast, Wright and colleagues report that cannabis use predicted increased depressive symptoms in both males and females, but increased anxiety symptoms and behavioral disinhibition were only found in females. Adolescent substance users have also been found to exhibit abnormalities in brain function, structure, and volume. However, given the nature of human studies, it is difficult to establish a causal link between early life exposure and the development of these conditions, especially as drug co-use is not often considered and may partially explain inconsistent findings noted in prior studies. Nicotine acts in the brain via the neuronal nicotinic acetylcholine receptors, which are ligand-gated ion channels expressed on both presynaptic and postsynaptic membranes. Rodent models have shown that adolescent nicotine exposure alone may lead to behavioral alterations during adulthood. For instance, in male and female rats, adolescent nicotine enhances nicotine reward and intake during adulthood.
Nicotine during adolescence has also been shown to increase depression-associated behaviors, decrease exploratory activity, and induce deficits in context conditioning to shock-associated cues in adult rats. However, in these studies, differences were not found with anxiety-associated behaviors, extinction of contextual conditioning, or cued fear responses. In mice, sex dependent effects have been noted, with adolescent nicotine consumption leading to decreased anxietyassociated behaviors in adult females, but not males. With regard to cannabinoids, Δ9- tetrahydrocannabinol has been classified as the main psychoactive component in cannabis and exerts its actions on cannabinoid 1 and cannabinoid 2 receptors in the brain and periphery. Differential patterns of expression for these receptors are found across adolescent development and between males and females, and notably CB1 receptors exhibit the highest level of expression during the developmental period of mid-adolescence. Following THC administration in adolescence, adult female, but not male, rats exhibit depression-associated behaviors, but no changes in anxiety-associated or general locomotor behaviors were observed. Interestingly, the depression-associated behavioral effects found in females were paralleled by significantly reduced CB1 receptor expression and activity in the amygdala, ventral tegmental area and nucleus accumbens, whereas similar changes were not found in the ventral tegmental area and nucleus accumbens of males. Further, administration of WIN 55,212–2, a CB1 and CB2 specific agonist, during adolescence has similarly been shown to increase depressive-like behaviors, as well as palatable food intake, during adulthood in male rats. Together, these prior findings demonstrate that early life exposure to either nicotine or cannabinoid agonists alone can alter later affective and cognitive function, which introduces the possibility of potential synergistic or opposing effects under co-use conditions.In the current studies, we sought to examine whether nicotine and cannabinoid co-exposure during mid-adolescence would result in altered affective and reward-seeking behavior during adulthood. While prior studies have examined each drug and/or behavioral measure independently, the current investigations represent the first study of a co-exposure condition, which is commonly found in human subjects, and the resulting effects on multiple cognitive and affective measures. To this end, adolescent mice were exposed to the cannabinoid receptor agonist, WIN55,212–2, and/or nicotine and then assessed for cognitive,vertical grow anxiety-related and depression related behaviors during adulthood. Drug exposure occurred during postnatal day 38–49, which corresponds to mid-adolescence in rodents or ~13–17 years of age in humans. Based on prior evidence of differential responses for males and females with drug-related effects and baseline receptor expression, male and female mice were examined in a within sex manner. Further, given that significant differences were found in behavioral measures at the moderate dose of the cannabinoid agonist, a second study was then conducted to examine whether these effects would be maintained with a lower dose of the cannabinoid agonist. Together, these studies provide evidence that adolescent drug exposure alters affective and reward-related behaviors during adulthood in a sex- and drug-dependent manner.Male and female wildtype C57BL/6J mice were derived from breeders in our laboratory animal facilities. Mice were maintained in an environmentally controlled vivarium on a 12 h reversed light/dark cycle. Food and water were provided ad libitum until behavioral training commenced.
During food training, subjects were mildly food restricted to 85–90% of their free feeding body weight, and water was provided ad libitum. Following food training and the lever reversal task, food and water were again provided ad libitum for at least 5 days prior to subsequent behavioral assessments. All experiments were conducted in strict accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at the University of California, Irvine.On PND 70, subjects were mildly food restricted and trained to press a lever in an operant chamber for food pellets under a fixedratio 5, time out 20 s schedule of reinforcement. Each session was performed using 2 retractable levers . Completion of the response criteria on the active lever resulted in the delivery of a food pellet. Responses on the inactive lever were recorded but had no scheduled consequences. Once stable responding was achieved , the lever assignment was switched to examine cognitive flexibility. In the reversal task, the previous inactive lever became active, in that food pellets were earned in accordance with the established FR5TO20s schedule. In contrast, the previously active lever became inactive, in which responses were recorded but without scheduled consequence.Subjects were habituated to sucrose pellet consumption for 2 days prior to sucrose testing, during which time 60 mg of sucrose pellets was provided for each subject in the home cage. On the third day, subjects were individually examined in home cage conditions, but were single housed and provided 200 mg of total sucrose pellets in a dish. All subjects were maintained under ad libitum full food conditions, and thus were not food restricted during testing. Sucrose eaten was recorded at specified intervals by experimenters blinded to the group condition. At the end of each session, experimenters examined the cage for breakage or disintegration of sucrose pellets; this occurred on only a few occasions and in these instances, the remnant amount was calculated and included in the final mg amount of sucrose remaining. Mice were required to consume at least one 20mg sucrose pellet within the first 30-min time period for inclusion in the study.Given the growing incidence of nicotine and cannabis experimentation during adolescence, we sought to examine whether such exposure would lead to altered behavioral effects during adulthood. In these studies, we found that male adolescent exposure to a moderate dose of the cannabinoid receptor agonist, WIN55,212–2 , led to increased cognitive flexibility in a learning reversal task, decreased anxiety-associated behaviors, and increased natural reward consumption, but no differences in general locomotor or depression-related behavior. Interestingly, the co-exposure condition of both nicotine and the moderate dose of WIN led to similar behavioral profiles as WIN alone in these measures, suggesting that a potentiative or additive effect was not present for these behaviors. However, with regard to the number of lane crosses in the elevated plus maze, the nicotine and WIN co-exposure condition appeared to exert a counteractive effect on the WIN-induced increase in exploratory behavior at the moderate dose, suggesting an opposing effect with adolescent exposure to both drugs. With regard to females, the moderate dose of WIN induced a lower body weight during the adolescent period, but co-exposure with nicotine appeared to exert an opposing effect that resulted in no difference from the control condition.