Chemsex Drugs on the Rise

A Longitudinal Analysis of the Swiss HIV Cohort Study From 2007 to 2017

B Hampel; K Kusejko; RD Kouyos; J Böni; M Flepp; M Stöckle; A Conen; C Béguelin; P Künzler-Heule; D Nicca; AJ Schmidt; H Nguyen; J Delaloye; M Rougemont; E Bernasconi; A Rauch; HF Günthard; DL Braun; J Fehr


HIV Medicine. 2020;21(4):228-239. 

In This Article


Study Design

The SHCS, launched in 1988, is a prospective cohort study with ongoing enrolment of HIV-diagnosed individuals in Switzerland. Of the approximately 20 000 participants in total, around 40% are MSM.[14] It is estimated that about 84% of all HIV-positive MSM living in Switzerland participate in the SHCS.[15] Detailed clinical, laboratory, demographic and behavioural data are recorded at study entry and every 6 months thereafter. The SHCS collects data at seven centres in Zurich, Basel, Bern, Geneva, Lausanne, Lugano and Sankt Gallen, as well as from associated hospitals, clinics and collaborating physicians specializing in HIV care.

Study Population

All patients enrolled in the SHCS with at least one visit between 2007 and 2017 were included. For this study, MSM are defined as male participants reporting a homosexual or bisexual preference. All other participants were categorized as non-MSM.

Study Measurements

Data Collection. The SHCS questionnaire is completed every 6 months by a health care professional in an interview with the participant. Since 2007, all SHCS participants have been asked about RDU. Data on heroin, cocaine and cannabis use are collected in the form of binary variables. For other drugs, more detailed information in two free text fields, one for intravenous drugs and one for nonintravenous drugs, can be provided. We combined information on intravenous and nonintravenous drug use of the same substance and did not analyse the mechanism of administration separately. We included information about the mechanism of administration as well as the amount of clearly misclassified entries in the 'Data quality' section and Appendix 1.

Substances of Interest. A pattern search algorithm was applied to the text fields containing information about drug use, and all entries were categorized into one of 16 drug classes (see Table A1 and Appendix 1): chemsex drugs (GHB/GBL, methamphetamine, ketamine and mephedrone), other sex-enhancing drugs [cocaine, 3,4-methylenedioxymethamphetamine (XTC/MDMA), amyl nitrite and amphetamine] and other drugs [cannabis, heroin, benzodiazepines, opioids other than heroin, lysergic acid diethylamide (LSD) or other psychogenic drugs, methylphenidate, anabolic substances and other prescription drugs]. Entries that could not be categorized by the algorithm were assigned to one of the drug classes by the authors, if possible, or left out of the analysis. The results were then combined with the binary entries for cocaine, cannabis and heroin.

First, we performed an analysis of the overall use of all 16 drug classes. Secondly, the time trend for all chemsex drugs and other potentially sex-enhancing drugs (excluding the less harmful substance amyl nitrite) was analysed for all SHCS participants, as well as for MSM and non-MSM separately. For the comparison of participants who ever reported drug use with participants who never reported drug use during the whole follow-up period, we concentrated on the two most important chemsex drugs, GHB/GBL and methamphetamine, in addition to all other potentially sex-enhancing drugs (again excluding the less harmful substance amyl nitrite).

Association Between Drug use and Sociodemographic, Psychosocial and Behavioural Measures. All SHCS participants are screened every 2 years for HCV (MSM yearly) and every year for syphilis. Since 2000, all participants have routinely been asked whether they have had anal or vaginal sex with casual partners and whether condoms were used all the time. Since 2003, patients have routinely been asked how often they have missed a dose of antiretroviral therapy (ART). The level of self-reported ART adherence is classified as follows: the answers 'never' and 'once a month' are grouped under 'good adherence', the answers 'once per week' and 'once every second week' are grouped under 'medium adherence', and the answers 'more than once per week' and 'every day' are grouped under 'poor adherence'. These adherence categories correlate well with viral suppression and mortality.[16] In 2008, a binary variable for depression was introduced, which relies on patient self-reporting.[17] Lastly, three education categories are used: patients who did not finish any education or only 9 years of mandatory school education, patients who finished an apprenticeship or A levels, and those who completed higher education including university.

Statistical Analysis

Time Trend of Drug use. To analyse the time trend of drug use in the SHCS, we extracted the number of patients using the different drugs for each year from 2007 to 2017. In the event of more than one follow-up visit per year for the same patient, the information was combined; that is, we included the information on whether this patient reported drug use at least once during the year of interest. We used generalized estimating equations (R package 'geepack') to account for correlated responses from participants over time.

Association Between Drug use and Sociodemographic, Psychosocial and Behavioural Measures. We determined whether drug use among MSM was significantly associated with the level of education, condomless sex, depression, adherence to ART, detectable HIV viral load, and HCV or syphilis coinfection. In particular, we compared MSM who had used methamphetamine, GHB/GBL, cocaine, XTC/MDMA or amphetamine, respectively, at least once during the whole follow-up time with a control group. The control group consisted of MSM who, at all follow-up visits in the years 2007–2017, either reported no recreational drug use or only the use of cannabis or amyl nitrite, as these drugs are often considered less harmful.[18] We did not include prescription drugs in our analysis, as information regarding whether the substance was used as prescribed or was abused was unavailable. We analysed the above-mentioned factors using logistic regression, univariable as well as adjusted for the sociodemographic factors age, ethnicity and year of diagnosis. MSM reporting use of the substances of interest were compared against the control group with no drug use, respectively. Data analysis was performed with R (version 3.3.0).[19]

Data Quality Check. To assess the quality of our data and to elucidate potential flaws in the way the data were collected, we performed a descriptive analysis of three aspects of the data. First, we counted the number of different spellings of the same substance to assess potential ambiguities in the classification. Secondly, we counted the number of entries that could not be assigned to any of the 16 drug classes because of vague description of the substances. Thirdly, we counted the number of entries that were placed in the wrong category regarding the mechanism of administration, that is, intravenous versus nonintravenous. Additionally, we compared the data to a currently unpublished data set based on an online questionnaire taken by 109 SHCS participants for the Swiss HCVree trial, a substudy of the SHCS which aims to screen and treat all SHCS participants for HCV.[20]