Shisha Consumption and Presence of Cotinine in Saliva Samples among Students in Public Universities in Coastal Kenya

Background Despite the well-known adverse health effects of tobacco, shisha use among students in tertiary institutions remains a public health concern. In Kenya, the literature on status of shisha after the 2017 ban is scanty. This study sought to ascertain actual shisha use among university students along the coastal strip. Methods We investigated confirmed and self-reported shisha use. Using proportionate-to-size and snowball sampling methods, 380 respondents were enrolled from three universities. Sociodemographic characteristics and self-reported history of shisha use were documented using a participant-assisted questionnaire. Actual shisha use was determined qualitatively using 6 panel plus alcohol saliva test kit that detected cotinine use among other selected drugs. Results Of the 380 participants, 278 (73%) were males and their median (IQR) age was 22 (20–23) years. This study reports 29% current use based on testing positive for cotinine. Among those who reported current ever use of shisha, 19% tested positive for cotinine, respectively. In the multivariable analysis, being separated (adjusted risk ratio (aRR): 2.06 (95% CI: 1.45–2.94)) compared to being single and studying for a degree compared to a diploma (aRR: 1.32 (95% CI: 1.10–1.58)) were associated with cotinine positive. The 4th year of study (aRR: 1.68 (95% CI: 1.22–2.33)) compared to the 1st year and reported knowledge of shisha (aRR: 1.84 (95% CI: 1.18–2.87)) were associated with cotinine positive. Conclusion Nearly one-third of university students along the Kenyan coast are active shisha users. Saliva testing for cotinine is a more reliable method of reporting tobacco use. We recommend upscaling of health education, re-enforcement of the current ban on shisha consumption by concerned authorities, and saliva testing for cotinine while assessing current tobacco use.


Introduction
Te tobacco epidemic remains a critical public health problem as it accounts for eight million deaths per year with about 80% of current tobacco users living in sub-Saharan Africa [1,2].If no interventions are put in place, the number of deaths attributed to tobacco exposure is projected to double by the year 2030 in low-and middle-income countries [3].
Waterpipe tobacco smoking (WTS) also known as shisha is a rapidly emerging trend among young consumers [3,4] especially in the Middle East and North Africa [5] partly due to the aromatic favours in shisha [6].Most shisha consumers believe it is less harmful compared to other forms of smoked tobacco [7].Shisha smokers are likely to inhale more tobacco smoke than from a cigarette because of the large waterpipes used and longer exposure [8].Tobacco smoke generated from shisha has a number of chemicals that are known carcinogens and tumour promoters [9].Te process of shisha preparation and the structure of a shisha hooker have been described in our previous work [10].
Currently, research on shisha use is scanty as Kenya implemented a comprehensive ban on shisha including its use, import, manufacture, sale, ofer of sale, advertising, promotion, and distribution in 2017 [24].Despite this comprehensive ban on shisha and prohibition of drug use within premises of learning institutions [25], shisha consumption remains rampant in Mombasa as well as learning institutions.Prior to the ban, a national status on shisha and tobacco use report indicated 9.1% of Kenyans were consuming tobacco products [26].
Compared to the MENA region, shisha use among university students in SSA is less explored, hence the need to carry out this study.Moreover, the literature available for both MENA and SSA is based on self-reported history that is prone to social desirability and memory recall biases.Additionally, in Kenya, the literature on status of shisha after the 2017 ban is scanty.It is for this reason we sought to ascertain actual shisha use among university students along the coastal strip.

Materials and Methods
2.1.Setting, Design, and Study Participants.Tis was part of a larger cross-sectional descriptive study that sought to assess the extent of substance among university students along the coastal strip of Kenya between 2018 and 2019.We enrolled students from three public universities, namely, Technical University of Mombasa (Main Campus), Moi University (Coast Campus), and Kenyatta University (Mombasa Campus), as summarized in Figure 1.

Ethical Considerations.
Te protocol for this study was ethically reviewed and approved by Pwani University Ethical Review Committee (ERC/Msc/010/2018).Participant enrollment was voluntary and through written consent.Prior to consenting, participants were educated on the objectives of the study and their right to voluntary participation as well as withdrawal from the study.Enrollment was carried out within university health clinics that provided privacy and confdentiality.Te participant-assisted questionnaires were coded to avoid use of names.Both hard and soft versions were stored and have restricted access.

Variables and Data
Source.Te exposures were sociodemographics, awareness level on shisha, and use of other tobacco products.Te study outcomes were reported shisha use status categorized into three levels as ever use, current use, and never use.Te saliva sample tested for shisha use was reported as a binary outcome, either positive or negative.

Sample Size Determination and Sampling.
Te study enrolled 384 participants using N � (1.96) 2 × 0.5(1-0.5)/0.052 � 384 [27].Our attribute under study (P) was 0.5 since at the time of enrollment, half of the residents of Mombasa County had reported a lifetime prevalence of at least one substance [28].At the time of the study period, the total population of TUM, KU, and Moi University was 2 Global Health, Epidemiology and Genomics 16,000, 5000, and 3000, respectively.Terefore, proportion to size sampling based on the ratio of 16 : 5 : 4 was used to calculate the representative sampling size from each university to make the total sample size of 384 participants.Te snowball sampling method was used to enroll the participants.Tese two sampling methods have been used elsewhere in similar subpopulation [29,30].Although snowball sampling method is a nonprobability sampling method, in research targeting difcult-to-reach population such as those using illegal drugs like shisha, it is considered irreplaceable and necessary sampling method.
Students were informed about the study through their WhatsApp group.Tis provided a fair chance of voluntary participation into the study.Each participant was verifed by providing a valid university identity card.To ensure confdentiality, enrollment was carried in an enclosed consultation room.Participant-assisted questionnaire was administered by a trained research assistant.Upon successful recruitment in the study, each participant was requested to tell someone from their class.Out of the 384 participants, one had invalid results and declined to be tested again; two declined to consent to saliva testing; and one participant-assisted questionnaire was incomplete.Te four participants were, therefore, excluded in the fnal analysis and had a total of 380 participants distributed as follows: 220 (58%); 100 (26%); and 60 (16%) from TUM, KU, and Moi universities, respectively.

Data Collection Tools and Procedures.
A participantassisted questionnaire was used to document social demographic characteristics, risk awareness level on shisha, and self-reported use of shisha.Te tool was pretested at TUM, Kwale Campus.Confrmed drug use was carried out using 6 panel plus alcohol saliva rapid test kits as per manufacturer's instructions.Te test kit is an immune-chromatographic assay that uses monoclonal antibodycoated gold particles.Te test kit comprises of mouth swab sponge and a test cube.Te mouth swab sponge was placed by a research assistant on the tongue or near the cheek to soak it in saliva until saturated (when the indicator strip appeared red).Te saturated mouth swab sponge was placed in the test cube and closed tightly.Results were read after fve minutes.Te appearance of one red line in the control meant a positive result while the appearance of two lines (one at a control panel and the other at a test panel meant negative results).When no red line appeared or one red line appeared at the test panel, it meant the results were invalid and the participant was requested to consent to a repeat test.Te test device detected the presence of cotinine (a metabolite of nicotine) alongside amphetamines, Cannabis, benzodiazepines, opiates, cocaine, and alcohol.

Statistical Methods.
Study data were collected using a paper questionnaire and later entered into the Epidata database.We did not assume missing data were at random; for any missing data, we added an extra category (unknown/ missing) and included it in the analysis.
Chi-square test of association or Fisher's exact test was used to compare the three levels of reported shisha use and diferent exposures.Using the saliva test for cotinine as the gold standard, we calculated the sensitivity and specifcity of reporting currently using shisha.Combining those who reported currently and ever used shisha as one group (shisha use), we also calculated their sensitivity and specifcity.To identify factors associated with testing positive for cotinine, we used a multilevel log-binomial regression analysis with the recruiting university as a random intercept (to account for inter-university variation).We used backward stepwise approach to select factors to include in the multivariable model.Te approach starts with a full (saturated) model and

Informed consent process
Filling in of participant assisted questionnaire

Saliva testing
Inform a classmate about the study Global Health, Epidemiology and Genomics only retains factors with a P value <0.1.As sensitivity analysis, we categorized reported shisha use into two levels: (a) no shisha use and (b) shisha use (currently using shisha plus the ever used).We similarly used multilevel logbinomial regression analysis to identify factors associated with shisha use.Te regression coefcients were logtransformed and reported as risk ratios and their respective 95% confdence intervals.Statistical signifcance was at α < 0.05.Statistical analysis was conducted using STATA version 17.0 (College Station, Texas 77845, USA).

4
Global Health, Epidemiology and Genomics    Global Health, Epidemiology and Genomics

Discussion
Tis study reports 29% current shisha use based on testing positive for cotinine.Te sensitivity of those who self-reported currently using shisha was about 68%, meaning more than two-thirds of them tested positive for cotinine while approximately 74% of those self-reported not to currently using shisha tested negative (specifcity).Tis highlights the importance of saliva testing because relying on self-reporting alone would have misclassifed 26% students as not using shisha, yet traces of cotinine were detected in their saliva.Globally, assessment of variation between self-reported and performance of saliva testing with sensitivity and specifcity values favours saliva testing as a more accurate tool for reporting tobacco use.For example, in Poland, about 5% of the participants who had been misclassifed as nonsmokers based on self-reported current tobacco use were classifed as smokers based on the cotinine cutof value [31].In India, self-reports among youths aged between 10 and 19 year had a low sensitivity (36.3%) and a positive predictive value of 72.6% [32].It is therefore important to validate selfreports on tobacco use using biochemical markers presents in body fuids like saliva.
Tese results on current use of cotinine deviate from fndings of a previous in the same region and similar subpopulation whereby 59% tested positive for cotinine [33].Such a deviation was also observed in the study in Nairobi, Kenya, whereby 21.5% of the participants self-reported current shisha use [21].A positive cotinine test among those who self-reported to have never used shisha could be due to poor memory recall, social desirability, passive smoking, or third-hand exposure.Both passive smoking also known as second-hand smoking and third-hand exposure are detrimental to human health [34,35].
In this study, male students were more likely to be currently using shisha than their female counterparts; however, the diference was not statistically signifcant.Tese fndings deviate from the results of a study at Jazan University where the prevalence rate of WTS was 34.0% and signifcantly higher in males (42.5%) than females at 27.0% [36].Te peak age group for shisha use was between 22 and 24 years.Tis is a deviation from a peak age of 25-29 in Rwanda [20], 21-25 in South Africa [22], and ages of initiation of 18-21 years among students in Jordan [37].Participants of Christian faith were more likely to consume shisha.Such a high prevalence has previously been reported Global Health, Epidemiology and Genomics among students in two Kenyan universities [21,33].However, the number of Muslim students enrolled in both studies was extremely low and this could have caused a bias.
Based on current shisha use, more participants in the third year of study were more likely to use shisha by virtue of testing positive for cotinine.Such an incremental prevalence that is proportional to year of study has been reported among medical students in Istanbul, Türkiye, and in a South African medical school [38,39].Initiation is more likely to take place in frst and second years of study by senior students who are already habitual users.Peer pressure, need to demonstrate adulthood, and academic stress could be the main drivers of shisha use in third year.Nonetheless, irrespective of the year of study, low cotinine test positivity rates were observed among participants from Moi University (Figure 2).Tis could be due to low number of participants or absence of environmental cues that trigger shisha use as discussed in the next paragraph.Participants from business-related studies were more likely to report current of shisha.Similar fndings were reported in Rwanda [20] and Bangladesh [40].Students enrolled in businessrelated studies are likely to use shisha due to idleness as the courses ofered are relatively lighter.
Participants with a moderate monthly stipend of US$1260 to 1890 were more likely to use shisha.Similar fndings have been reported in Taibah University in the KSA and Ankara in Türkiye [41,42].In general, students with higher socioeconomic status and having higher stipends per week had increased odds for shisha usage unlike other drugs where poverty is a risk factor.Participants from TUM were more likely to use shisha than those from KU and Moi universities.Tis could be due to the overrepresentation of participants enrolled from TUM or the presence of environmental cues within TUM.Environmental cues associated with recreational drugs may trigger cravings or a relapse, thus infuencing the risk of shisha use [43].
Regarding awareness level, two-thirds of the participants knew the efects of shisha including those currently using shisha.Tese fndings deviate from most studies in the MENA region where most consumers lacked knowledge on the efects of shisha [7,11].Despite knowing the negative health efects of nicotine, the participants were not deterred from consuming shisha.Tis could be due to the fact that most shisha manufacturers provide deceptive information [44].Consequently there is need to upscale health education on the negative efects of shisha use in the wake of misinformation about commercial tobacco products.Most shisha consumers were most likely to consume shisha at the beginning of the semester partly due to free time or idleness.In Malaysia, a study reported shisha consumption among university students as a favourite pastime activity [45].
Findings on association between being married mirror the results of a study in the Qassim Region of KSA [18].Tis could be due to marital stress or fnancial demands that come with marriage and hence the use of shisha to relax.Ironically, being separated was also a predictor of shisha use.Tis could be attributed to life post-separation which might be characterised by loneliness.Loneliness is a predictor of waterpipe use [46].Despite having fewer Muslim participants, being a Muslim predicted shisha use.Tis can be attributed to Arabic culture since shisha is trendier in MENA and Eastern Mediterranean regions [5,46,47].Relationship between being in year one to three of study and shisha use could be as a result of tobacco being a gateway drug [48].
Te strength of this study is anchored on confrmed shisha consumption using oral fuids which is more reliable than self-reported shisha use.Te fndings of this study have shed more light on status of shisha use after the 2017 ban.Te limitation of this study is the cross-sectional design which cannot establish evidence for temporal relationship between the exposures and shisha use.Additionally, use of the snowball sampling method could have introduced a selection bias.Future studies to consider using respondent driven sampling; use of hair samples and environmental toxicology to detect shisha use up to a window period of six months retrospectively while the later can detect third-hand exposure.

Conclusion
Nearly one-third of university students along the Kenyan coast are active shisha users.Saliva testing for cotinine is a more reliable method of reporting tobacco use.We recommend upscaling of health education, re-enforcement of the current ban on shisha consumption by concerned authorities, and saliva testing for cotinine while assessing current tobacco use.

Figure 1 :
Figure 1: Flowchart of the study on shisha use among students in public universities in Coastal Kenya between 2018 and 2019.

Figure 2 :
Figure 2: Proportion who tested positive for cotinine across the year of study stratifed by the three public universities in Coastal Kenya in 2018-2019.

Table 1 :
Self-reported and confrmed shisha use among students in public universities in Coastal Kenya in 2018-2019.

Table 2 :
Sociodemographic characteristics and cotinine test results based on saliva analysis among students in public universities in Coastal Kenya between 2018 and 2019.
* P values from chi-square/Fisher's exact test; all results are expressed as frequency and proportion;

Table 3 :
Awareness levels on shisha use by students in public universities in Coastal Kenya between 2018 and 2019.

Table 4 :
Univariate and multivariable analysis of factors associated with cotinine positive among students in public universities in Coastal Kenya between 2018 and 2019.

Table 4 :
Continued.not selected for inclusion in the multivariable regression model, risk ratios from a log-binomial regression models, US dollars exchange rate was KES 125.9353 per one US dollar in January to February 2023 (https://www.centralbank.go.ke/rates/forex-exchange-rates/).Statistical signifcance at α < 0.05.

Table 5 :
Univariate and multivariable analysis of factors associated with reported shisha use by students in public universities in Coastal Kenya between 2018 and 2019.

Table 5 :
Continued.not selected for inclusion in the multivariable regression model, risk ratios from a log-binomial regression models, US dollars exchange rate was KES 125.9353 per one US dollar in January to February 2023 (https://www.centralbank.go.ke/rates/forex-exchange-rates/).