The Intersection between Spliff Usage, Tobacco Smoking, and Having the First Joint after Waking

Cannabis users who are also tobacco smokers are more likely to exhibit cannabis dependence and psychosocial problems. However, there has been minimal research around various cannabis and tobacco mixing (spliff usage) behaviours and likeliness to smoke the first joint within an hour of waking, known colloquially as wake and bake. The time of first joint and spliff usage may be related as they are associated with the intersection of tobacco and cannabis use. Compared to non-morning cannabis users, morning users reported significantly more cannabis-related problems. Through a survey of US cannabis users, we test the association between various cannabis and spliff use behaviours and likeliness to smoke the first joint within an hour of waking. Compared to those who smoked tobacco and used spliffs, the following spliff use behaviour groups were less likely to have their first joint within 60 minutes after waking: those who smoked tobacco and used spliffs (95%CI: 0.605–0.988); those who never smoked tobacco and did not use spliffs (95%CI: 0.489–0.892); those who never smoked tobacco and used spliffs (95%CI:0.022–0.915). We provided possible explanations for our results and suggested further research to better understand findings, important given expanding US cannabis markets.

values. We controlled for demographic characteristics, tobacco use in the last year, time of last cannabis joint, amount of cannabis used per session, number of hours stoned per session and number of days cannabis was used in the last year. We included how long before bed participants had their last joint, as cannabis is sometimes used for sleep initiation and this form of use may contribute to cannabis dependence 34 . Frequency and quantity used per session are associated with problematic cannabis use 35 . We included number of days cannabis was used in the last 12 months and amount of cannabis used per session. Time spent intoxicated on cannabis is a marker of problematic use 36 and we included a measure of number of hours spent stoned in a session.
Odds ratios (ORs), adjusted odds ratios (aORs), 95% confidence intervals (95% CI) and p-values were reported. Participants were not required to answer every question, resulting in some missing data. Given the structured sequence of questions, missing values for mixing tobacco with cannabis meant that participants had not used a spliff e.g. if someone indicated they did not use cannabis, the mixing tobacco with cannabis (spliff use) value would be missing. Thus, we coded missing values for mixing tobacco with cannabis (spliff use) as Never. When accounting for questions participants were not required to answer, the percentage of missing values for www.nature.com/scientificreports www.nature.com/scientificreports/ all variables did not exceed 20% (see Supplementary Table I). Multiple imputation was utilized to appropriately handle missing values, assumed to be missing at random. Given the number of categorical variables, we used the predictive mean matching technique 37 with only the variables described in this study. Similar analyses were run on the imputed dataset (see Supplementary Table II). We generated 10 imputation data sets. Statistical tests were performed on original and imputed data sets to determine the extent of result convergence 38 . All analysis was conducted using R with the following packages: dplyr, stargazer, plyr, lmtest, multiwayvcov, sandwich, mice 37,39-46 .

Results
Sample. From November 2016 to January 2017, a total of 10183 responses were recorded in the US. 8345 (82%) participants reported cannabis use in the past year. A further 1955 records were excluded due to missing data. The remaining 6390 respondents formed the sample for analysis. Males accounted for 75.48% of the sample, with a median age of 23 (interquartile range (IQR): 19-32, Range: 16-79, see Table 1).

Demographic characteristics.
Of those who reported cannabis use in the past year, most (78%) reported consuming their first joint more than an hour after waking and most (78%) tended not to use spliffs (see Table 1). Participants used cannabis for a median of 250 days in the last year (almost daily), with 0.50 grams the median for use per session. Participants spent a median of four hours a day stoned when cannabis was used. Those who had their last joint just before bed spent a median of six hours a day stoned. About half the sample had their last joint 1-2 hours before bed. When comparing cannabis use behaviours across spliff use groups, there were clear differences in gender distributions (see Table 2). For those who smoke tobacco and used spliffs, and never smoke tobacco and used spliffs, about half were in the 16-20 age group. For all other groups and the broader sample, the 16-20 group was a relatively small proportion. Distributions across all other variables were relatively similar across groups and mirrored the broader sample.
Consuming first joint within an hour after waking and spliff use. No statistically significant associations were found for the logistic regression with or without controls, for the relationship between spliff use (Never, Yes) and smoking the first joint within an hour after waking. Table 3 indicates the results of the logistic regression, with and without controls, regarding consuming the first joint within an hour after waking, and various spliff use patterns. When accounting for controls, compared to those who smoked tobacco and did not use spliffs, the following groups were less likely to have their first joint within 60 minutes after waking: those who smoked tobacco and did use spliffs (aOR = 0.80, p = 0.02), those who never smoked tobacco and did not use spliffs (aOR = 0.69, p = 0.00), those who never smoked tobacco and did use spliffs (aOR = 0.47, p = 0.00). Those who never smoked tobacco and did use spliffs had the lowest probability of smoking within the first hour and those who smoked tobacco but did not use spliffs had the highest probability (see Fig. 1). Results of logistic regression analysis run on the original and imputed data converged for all explanatory variables and categories (see Supplementary Table II).

Discussion
We sought to test the association between various spliff usage behaviours and likeliness to smoke the first joint within an hour of waking, among US cannabis users. Accounting for controls, compared to those who smoked tobacco and did not use spliffs, the following were less likely to have their first joint within an hour after waking: those who smoked tobacco and used spliffs, those who never smoked tobacco and did not use spliffs, those who never smoked tobacco and used spliffs.
There are a range of explanations for our results. We explore two possible explanations, in line with the limitations of our data. Harm reduction, of a life-functioning variant instead of a pulmonary/respiratory variant 47 , may explain why respondents who smoked tobacco and used spliffs, did not want to have their first joint within an hour of waking. Some participants may be aware of the harms of tobacco smoking combined with spliff use and thus seek to reduce engagement in another harmful behaviour; waking and baking. Another explanation may be demographic differences across spliff usage groups. We indicate proportionately more younger participants in certain spliff usage groups, possibly related to our outcome variable. Similarly, cannabis and tobacco users have differing outcomes compared to those who use only cannabis 12 . Qualitative research around specific demographic groups and tobacco and cannabis use behaviours may provide further insight.

Limitations.
We conducted likely the largest US study testing the association between various spliff usage behaviours and having the first joint within an hour after waking. This research design has costs and benefits, such as population level 29,48,49 reliability and validity. When data are limited, online surveys may be valid. Comparable probability sampling and ethnographic data may be key to increase validity of our findings 50 . The age and sex distributions of cannabis users who completed the GDS were similar to demographic distributions in a household survey across Australia, the US, and Switzerland 29 . The GDS is therefore an efficient means of getting a gradated insight around stigmatised behaviours as long as the survey is not used to determine population-level drug prevalence 29 .
We did not have information on how nicotine dependence may have influenced the results; frequency of tobacco use and time to first tobacco cigarette. This information may help explain our findings and future research can incorporate these survey items. As we used an online survey of US drug users, our sample was skewed toward younger participants. We used age and sex as controls, but other covariates such as sexuality, urban/non-urban residence and recruitment mode were not included. Not everyone in the sample reported smoking joints and future studies can exclusively recruit joint smokers. Our definition of spliffs did not include blunts, which can be  www.nature.com/scientificreports www.nature.com/scientificreports/ detailed in future research. The survey item did not explicitly refer to spliffs and it is possible that some respondents thought the item indicated other ways of mixing cannabis with tobacco. We will explicitly refer to spliffs in future survey iterations. It is possible that participants who reported not smoking tobacco but use spliffs mistakenly reported consuming tobacco on one question but not the other. We were not able to control for such effects but will include corrective mechanisms in future survey iterations. Those who had their last joint just before bed had a greater median time spent stoned compared to the larger sample. We did not conduct analysis with time of last joint, which may be marker of problematic use. We handled missing data with multiple imputation. While results from original and imputed datasets converged, a reduced rate of missing data would increase reliability of findings. concluding statement. Accounting for controls, in the US, compared to those who smoked tobacco and did not use spliffs, those with the following spliff usage behaviours were less likely to have their first joint within 60 minutes of waking: those who smoked tobacco and used spliffs; those who never smoked tobacco and did not use spliffs; those who never smoked tobacco and used spliffs. We provided some possible explanations for our results and suggested further research to better understand findings. Overall, we shed light on time of first joint and spliff usage behaviours, important given expanding US cannabis markets.