Using time series analysis, this is the first study to examine trends in hospital presentations for severe acute toxicity involving SCRA exposure before and after the implementation of the PSA to (i) use analytical confirmation of SCRA exposure and (ii) collect data from several hospitals from different regions across the UK.

Overall, SCRAs were detected in ~36% of patients presenting to participating IONA hospitals with severe acute toxicity suspected to involve NPS between July 2015 and December 2019. Across this period, we found no clear evidence of an upward or downward trend in the number of hospital presentations involving SCRA exposure before or after the implementation of the PSA or before or after the amendment to the MDA. There was also no clear evidence of a significant trend in the number of hospital presentations involving exposure to non‐SCRA drugs before or after either of these two policy changes. Interestingly, there was evidence of seasonal trends in the number of hospital presentations involving both SCRA and non‐SCRA exposures, with presentations increasing over summer months (i.e., May to September). Although not explored further, this is likely to reflect seasonal variability in the prevalence of drug use [29] and/or time available for participant recruitment at participating sites.

Our findings contrast with previous studies reporting increasing indicators of SCRA related mortality and morbidity since the PSA [4, 16, 17, 18, 23]. Increasing trends in SCRA related mortality reported by the ONS and NPSAD [4, 17, 18] may be a reflection of more detailed forensic analysis of drug‐related deaths and improvements in methods for detecting SCRA over time. However, in the current study, analytical methods for detecting SCRA in patient samples were consistent throughout the study period, but with reference libraries updated to include new compounds as they emerge. Our findings also contrast with data from an individual (London) hospital that showed that the number of presentations for acute toxicity involving patient‐reported SCRA exposure had increased in the year after the implementation of the PSA [23]. Those data are confined to a single region and are only inclusive of the year after the PSA; however, this may reflect the lack of consistency between patient‐reported and analytically confirmed exposure details. Previously, Abouchedid et al. [24] reported that SCRAs were reported by only 50% of patients in whom they were detected, whereas Tebo et al. [25] reported that SCRAs were reported (or suspected by clinicians), but not detected in ~45% of patients. In the current study, SCRA use was reported (or suspected by clinicians) in ~65% of patients with confirmed exposure and ~34% of those without. Although there is a possibility that novel SCRAs not detectable by current analytical methods may be present in patient samples, the likelihood of this is small because analytical methods used here can account for the emergence of new compounds and indicate the presence of previously unidentified ones. It is more likely that discrepancies between drugs detected and those reported by patients or suspected by clinicians reflect the mis‐sale or contamination of drug products, the prolonged time between exposure and hospital presentation/sample collection and/or the misattribution of other drugs' effects to SCRA use. It might also reflect potential limitations of self‐report measures to correctly identify SCRAs using suitable terminology understood by clinicians and the full drug using population [30]. In any case, these findings highlight the importance of analytical confirmation when investigating SCRA exposure.

The prevalence of SCRA compounds detected here—with 5F‐MDMB‐PINACA, AMB‐FUBINACA and MDMB‐CHMICA being predominant—are similar to previous reports in the UK and elsewhere [18, 24, 31, 32]. There was notable variability in the prevalence of these compounds across the study period, and consistent with a previous analysis of compounds presents in seized SCRA products [33] this appears to coincide with changes to legislative controls of SCRA compounds in China in 2018, as well scheduling under the MDA in the UK in 2016. In the majority of SCRA exposure cases, use of other substances were also detected; most common were opioids, benzodiazepines and cocaine. Also consistent with previous reports [18, 24, 25, 32], in the majority of cases, multiple different SCRAs were detected. Although this may be because of the use of separate products, SCRA products often contain multiple compounds in varying concentrations [16, 33], and it is possible that co‐administration with SCRA and/or other drugs may influence toxicity. Because this appears to be common among people who use SCRAs, enhanced toxicovigilance within clinical settings is necessary to help understand the interaction effects between different SCRAs and other commonly co‐administered drugs and improve the treatment response.

This study has important limitations. First, with a total of 224 confirmed SCRA cases across a 54‐month period, these data are only likely to represent a small number of SCRA related hospital presentations, and accurately estimating trends in data of this size can be problematic. Additionally, data were only collected for 11‐months before the implementation of the PSA, and we were not able to estimate potential anticipatory effects occurring in the period between the PSA being announced and implemented. It has been reported that many outlets held ‘fire sales’, in which remaining stocks of SCRAs and other NPS were sold at heavily discounted prices before the ban took effect [15], and this may have resulted in increased hospital presentations. Second, there were changes in the hospitals collecting data during the study period. Although analyses were adjusted for the changing number of hospitals over time it is possible that regional variability in patterns of drug use may have influenced the trends in SCRA (and non‐SCRA) detections as sites became active and inactive. Third, because of these inconsistencies in the data collection periods and difficulties in appropriately modelling potential anticipatory effects, we only examined changes in the trends within and not between the pre‐intervention and post‐intervention periods. For the same reasons, and because of high month‐to‐month variability in hospital admissions, we also did not attempt to interpret changes in the y‐intercept between these two periods, which may have provided an indication of any immediate changes in the number of hospital presentations occurring immediately after the PSA was implemented. Therefore, although these data can provide insight into the monthly trend in SCRA related hospital admissions since the implementation of PSA, it is not possible to comment on the policy's effect on pre‐policy hospital presentation levels. Fourth, although data were collected from 34 hospitals throughout the UK, hospital participation was voluntary and not determined by probability‐based sampling methods, therefore, these data cannot be considered nationally representative. Fifth, although we also examined trends in SCRA related hospital admissions before and after the amendment to MDA we did not explore potential impacts of international policy changes occurring within the study period (e.g. national control of several SCRA compounds in 2015 and 2018 in China) [11]. Finally, our analysis of patient samples could only determine whether or not SCRAs (and other drugs) were present, and we could not quantify concentrations or their clinical significance. It is possible that SCRAs present in analysed samples, especially those with prolonged elimination (including active metabolites), were used in earlier episodes of intoxication and may not have contributed to the toxic effects for which patients were seeking treatment. Nonetheless, these data provide an important indication of trends in the number of SCRA clinical exposures before and after implementation of the PSA.